2015-06-01 05:52:51 <fabled> duh
2015-06-01 09:41:15 <kaniini> ncopa: http://turtle.dereferenced.org/~kaniini/interfaces.conf.txt is what i have come up with lately
2015-06-01 09:43:34 <ncopa> how does it compare to openwrt?
2015-06-01 09:43:53 <ncopa> it it a completely new scheme or is it compatible with something else?
2015-06-01 09:46:05 <kaniini> ncopa: it is vyatta-like
2015-06-01 09:46:10 <ncopa> ok
2015-06-01 09:50:25 <kaniini> ncopa: the idea is to also have plugins which can extend it so that it can do things like set up VPNs automatically
2015-06-01 09:50:55 <kaniini> ncopa: and also that it tracks interfaces by MAC so that if the kernel reorders them it can still properly configure the box
2015-06-01 09:51:34 <ncopa> i like that
2015-06-01 09:52:03 <ncopa> i wonder if we could have some auto migrate from current ifupdown style config
2015-06-01 09:52:09 <kaniini> yes, it should be possible
2015-06-01 09:52:12 <kaniini> for most cases
2015-06-01 09:53:05 <ncopa> how does it compare to what systemd does?
2015-06-01 09:53:47 <barthalion> doesn't matter
2015-06-01 09:53:51 <barthalion> networkd is terrible :(
2015-06-01 09:54:11 <ncopa> ok :)
2015-06-01 09:54:16 <ncopa> not surprised
2015-06-01 09:54:22 <barthalion> I'm probably the biggest fan of systemd here and I cannot say anything good about networkd
2015-06-01 09:54:32 <barthalion> except it does work for very simple cases
2015-06-01 09:54:59 <barthalion> but it has so many quirks that I'd rather use networkmanager or commit a seppuku
2015-06-01 09:55:47 <ncopa> how about using something cisco-like?
2015-06-01 09:56:13 <ncopa> vyatta-like sounds reasonable though
2015-06-01 09:56:27 <fabled> ncopa, quagga is the cisco-like; and supports configuring addresses
2015-06-01 09:56:37 <fabled> though, it's not supporting tunnels, bond, vlans etc.
2015-06-01 09:56:56 <ncopa> would it be an idea use similar style for our interfaces config?
2015-06-01 09:57:13 <fabled> not necessarily
2015-06-01 09:57:19 <fabled> cisco-style config is a lot of cruft
2015-06-01 09:57:26 <fabled> and repetition
2015-06-01 09:57:36 <ncopa> i'm not saying vyatta-like is bad, im just trying to put the alternatives on the table
2015-06-01 09:57:40 <ncopa> so we can exclude them
2015-06-01 09:57:59 <ncopa> openwrt has its style too
2015-06-01 09:58:29 <fabled> why not use connman or similar
2015-06-01 09:58:40 <fabled> hmm... that probably depends on gtk/whatever
2015-06-01 09:58:50 <ncopa> dbus likely
2015-06-01 09:59:05 <barthalion> on gtk, but dbus probably
2015-06-01 09:59:16 <ncopa> i do think it might make sense to provide a dbus api
2015-06-01 09:59:19 <ncopa> dbus bridge
2015-06-01 09:59:37 <ncopa> but without depending on dbus to work at all
2015-06-01 09:59:44 <kaniini> well, dbus bridge could be done via plugin perhaps.
2015-06-01 09:59:49 <ncopa> yeah
2015-06-01 09:59:51 <ncopa> soemthing like that
2015-06-01 10:00:04 <fabled> kaniini, writing something to handle vlans, teams, bonds, bridging, device renaming, hotplug and address config is a lot of work
2015-06-01 10:00:25 <fabled> not sure if we want to go there
2015-06-01 10:00:36 <fabled> not to speak about wlans
2015-06-01 10:01:08 <ncopa> we could start with something simple and make it extendable with plugins
2015-06-01 10:01:37 <ncopa> i am slightly sceptical to make soemthing new (NIH syndrome?)
2015-06-01 10:01:52 <kaniini> well the only other thing that comes close is systemd-networkd
2015-06-01 10:02:01 <kaniini> which uses .ini-style configuration
2015-06-01 10:02:05 <kaniini> and is otherwise quite broken :P
2015-06-01 10:02:16 <ncopa> so i'd like a list of altnernatives, and a good reason to exclude them
2015-06-01 10:02:30 <fabled> is existing network/interfaces broken beyond fixing?
2015-06-01 10:03:03 <ncopa> yes, can we re-use existing format?
2015-06-01 10:03:06 <kaniini> fabled: in my opinion -- yes
2015-06-01 10:03:18 <kaniini> because for example, dual-stack is hard to set up properly with it
2015-06-01 10:03:29 <ncopa> multiple ip addresses
2015-06-01 10:03:40 <fabled> yes, it has problems
2015-06-01 10:03:48 <fabled> but i'm wondering if they can be fixed within the limits of the format
2015-06-01 10:03:51 <kaniini> the config grammar is also inconsistent
2015-06-01 10:04:15 <ncopa> kaniini: do you think you could add example if ppp interface
2015-06-01 10:04:18 <fabled> it's perfectly ok to have "inet eth0 inet static" and "inet eth0 inet6 static"
2015-06-01 10:04:29 <fabled> but then setting up vlans+bonding is painful
2015-06-01 10:04:42 <fabled> kunkku suggested having "iface eth0 link" or similar
2015-06-01 10:04:55 <fabled> earlier... for the device stacking portions
2015-06-01 10:05:13 <fabled> multiple addresses is trickier
2015-06-01 10:05:32 <kaniini> note that changing the syntax removes debian compatibility anyway
2015-06-01 10:05:51 <kaniini> there exists right now several well-defined parsers for the syntax i mocked up
2015-06-01 10:05:52 <fabled> i'd rather do that, than write something from scratch
2015-06-01 10:06:04 <fabled> but that's just me
2015-06-01 10:06:08 <kaniini> so writing from scratch is not really a thing
2015-06-01 10:06:23 <fabled> it's not the parser i'm worried
2015-06-01 10:06:33 <fabled> it's all the glue to map it to netlink / iproute calls
2015-06-01 10:06:45 <fabled> it's a considerable amount of code
2015-06-01 10:07:30 <kaniini> hmm, looking at systemd-networkd it does not look that involved
2015-06-01 10:08:21 <barthalion> well, networkd is pretty much underdeveloped now…
2015-06-01 10:08:40 <barthalion> but that can be probably said about 3/4 of new features in systemd
2015-06-01 10:08:49 <barthalion> what I mean is it doesn't do any magic yet
2015-06-01 10:09:55 <kaniini> yes, but my point is that the glue to set up the interfaces / routes is only a small part of sd-networkd
2015-06-01 10:10:11 <barthalion> yeah
2015-06-01 10:11:30 <fabled> then again, writing it in lua or similar, that maps it to iproute2 calls, should not be awfully complicated
2015-06-01 10:11:30 <kaniini> anyway, i hope to have a working test version over the summer
2015-06-01 10:12:58 <kaniini> also, replacing pkg-config was said to be difficult but it got done so :p
2015-06-01 10:14:14 <ncopa> woudl also be nice to have a look at open/net/freebsd
2015-06-01 10:14:16 <ncopa> to compare
2015-06-01 10:14:40 <ncopa> i like the "start when-hotplugged;" stuff
2015-06-01 10:14:54 <kaniini> open/net/freebsd use rc.network which basically is even more limited than what we have now
2015-06-01 10:15:48 <ncopa> vlan, ppp and openvswitch examples would be nice
2015-06-01 10:16:10 <kaniini> ppp/openvswitch we do not presently support do we
2015-06-01 10:16:21 <ncopa> we actually do
2015-06-01 10:16:27 <ncopa> ppp we do atleast
2015-06-01 10:16:27 <kaniini> in /etc/network/interfaces ?
2015-06-01 10:16:29 <ncopa> yes
2015-06-01 10:16:36 <ncopa> the openvswitch is hackish
2015-06-01 10:16:41 <ncopa> works almost :)
2015-06-01 10:17:35 <ncopa> auto ppp0
2015-06-01 10:17:35 <ncopa> iface ppp0 inet ppp
2015-06-01 10:17:35 <ncopa>         provider E378
2015-06-01 10:17:45 <kaniini> not sure yet, i was planning to start simple.  but that doesn't seem that complex to do.
2015-06-01 10:18:08 <ncopa> i mean, examples of how the config could look like
2015-06-01 10:18:18 <kaniini> probably something like
2015-06-01 10:18:19 <ncopa> the actual implementation can ofc be done much later
2015-06-01 10:18:36 <ncopa> vlan tags are important though
2015-06-01 10:18:43 <kaniini> interface "ppp0" { type ppp; start when-requested; provider "E378"; };
2015-06-01 10:18:47 <kaniini> which would use pppd
2015-06-01 10:18:59 <ncopa> yeah
2015-06-01 10:19:07 <kaniini> perhaps later adding ppp support could be done (i know pppd is not the greatest software :))
2015-06-01 10:19:46 <kaniini> vlans would be like
2015-06-01 10:20:10 <kaniini> interface "eth0.X" { type vlan; master "eth0"; tag 1234; };
2015-06-01 10:59:04 <ncopa> http://wiki.openwrt.org/doc/uci/network
2015-06-01 11:19:05 <[xming]> I never like the busybox' syntax
2015-06-01 11:19:59 <[xming]> kaniini's is better except for the semi colomns
2015-06-01 11:20:41 <[xming]> gentoo's style isn't bad either, is that gentoo specific or openrc?
2015-06-01 11:22:10 <[xming]> maybe json, yaml, .. compliant to make it easier to parse in any lang?
2015-06-01 11:23:21 <[xming]> Or Cisco ios a like so we can attract CCIEs :p
2015-06-01 12:59:36 <ahills> ncopa / kaniini: reason not to use connman, I've used it, the interface is very unwieldy, and if you're on wifi, it's not any better than just manually using wpa_supplicant (in fact it doesn't work in many cases that wpa_supplicant does)
2015-06-01 13:00:09 <ahills> I think yaml is a bad format and we should avoid it; json is very annoying to type manually
2015-06-01 13:00:29 <ahills> kaniini's style is the same as e.g. nginx, which is very common, easy to read and type
2015-06-01 13:08:13 <barthalion> also we are C guys, we love semicolons
2015-06-01 13:10:19 <ahills> haha
2015-06-01 13:10:34 <ahills> I think rigid syntax is good for system configuration personally
2015-06-01 13:10:43 <ncopa> :)
2015-06-01 13:10:50 <ahills> but that might be because C has turned my mind into cement
2015-06-01 13:28:59 <ttilley> 
ACTION <3 yaml

2015-06-01 13:30:18 <ttilley> but i started out as a ruby guy, so my world view is a bit different
2015-06-01 13:31:17 <ttilley> also -love- coffeescript, which has a lot of syntax that is effectively yaml
2015-06-01 13:34:50 <ahills> how can you even handle coffeescript scoping?
2015-06-01 13:35:05 <ahills> it makes even basic use of the language excruciatingly painful
2015-06-01 13:37:12 <[xming]> I can't stand ;
2015-06-01 13:37:56 <[xming]> and since compilers know that I missed a ; it should just stop whining :D
2015-06-01 13:39:16 <barthalion> then shit like in javascript happens
2015-06-01 13:39:30 <ahills> optional ;
2015-06-01 13:39:53 <barthalion> and you never know if ; has been placed in correct place
2015-06-01 14:06:19 <kaniini> javascript is best
2015-06-01 15:56:24 <jomat> Mh... adding an URL to a patch to source= doesn't work as expected :-(
2015-06-01 17:58:43 <tdtrask> ncopa: saw weird error when upgrading from 3.1 to 3.2 with lua-posixtz installed
2015-06-01 17:58:54 <tdtrask> created issue #4261
2015-06-02 06:51:38 <fabled> ttilley, re: gcc 5.1
2015-06-02 06:52:12 <fabled> yes, sounds like something we could do in edge; iirc, there was someone who had rebased our patches; was it you or someone else?
2015-06-02 06:52:26 <ncopa> we want gcc-5.1 asap
2015-06-02 06:52:42 <ncopa> i'd like to have a look at abuild and -dev packages first
2015-06-02 06:52:47 <ncopa> then look at gcc-5.1
2015-06-02 06:52:56 <ncopa> unless someone beats me to it
2015-06-02 07:23:19 <fabled> ncopa, i'll try to rebase gcc 5.1 patches then
2015-06-02 07:24:41 <fabled> i think many pr patches can be dropped
2015-06-02 07:24:44 <fabled> fixed upstream
2015-06-02 07:24:51 <ncopa> very well
2015-06-02 07:24:59 <fabled> remaining conflicts are
2015-06-02 07:25:10 <fabled> two of the gentoo pie/ssp patches; i'll check gentoo if they updated
2015-06-02 07:25:15 <ncopa> oh
2015-06-02 07:25:16 <fabled> two of my patches
2015-06-02 07:25:17 <ncopa> btw
2015-06-02 07:25:21 <fabled> and gcc49-cloog-dl.patch
2015-06-02 07:25:41 <ncopa> i got a twitter message bout our gcc .spec files
2015-06-02 07:25:50 <ncopa> the ones we picked from gentoo
2015-06-02 07:26:01 <fabled> oh?
2015-06-02 07:26:13 <fabled> i think i modified them slightly
2015-06-02 07:26:15 <ncopa> Hi @n_copa, we had a hard problem with using GCC with CGO caused by spec files you ship with the package. http://bit.ly/1JeH5tO
2015-06-02 07:26:26 <ncopa> apparently about gcc-go
2015-06-02 07:26:34 <ncopa> i didnt check out what that was about
2015-06-02 07:26:35 <ncopa> yet
2015-06-02 07:28:27 <fabled> cgo is a wrapper making C library bindings to Go iirc
2015-06-02 07:28:40 <ncopa> ok
2015-06-02 07:28:42 <fabled> sounds like it's incompatible with pie
2015-06-02 07:28:45 <ncopa> i might look at that later
2015-06-02 07:28:48 <ncopa> yes
2015-06-02 07:31:04 <fabled> i wonder why it's using -r
2015-06-02 07:31:33 <fabled> probably some sort of 'incremental' linking thing
2015-06-02 07:32:25 <fabled> i'll rename many of the gcc patch files, so we get them in order
2015-06-02 07:34:36 <ncopa> i should hurry up with abuild then
2015-06-02 07:38:36 <fabled> ok, seems cloog is no longer a dependency
2015-06-02 07:39:59 <fabled> and patch to make isl runtime optional
2015-06-02 07:40:06 <fabled> maybe we should take that
2015-06-02 07:40:14 <fabled> from fedora
2015-06-02 07:40:22 <ncopa> might be an idea yes
2015-06-02 07:40:34 <ncopa> hum
2015-06-02 07:40:50 <ncopa> but that also means that if isl runtime is not there, it will not be used
2015-06-02 07:40:58 <ncopa> and it will not be automatically pulled in?
2015-06-02 07:41:09 <ncopa> we might want add it manually to depends then?
2015-06-02 07:41:10 <fabled> yeah, we'd need to have it then explicitly in depends
2015-06-02 07:41:15 <fabled> dunno
2015-06-02 07:41:15 <ncopa> ok
2015-06-02 07:41:24 <fabled> let's hold that for now then
2015-06-02 07:42:58 <fabled> gentoo adds that too, though
2015-06-02 07:44:34 <ncopa> gentoo might need it
2015-06-02 07:44:44 <ncopa> gentoo need gcc to be able to upgrade any package
2015-06-02 07:45:09 <ncopa> i think the idea is to simplify isl upgrade
2015-06-02 07:45:25 <ncopa> you upgrade isl with new abi
2015-06-02 07:45:34 <fabled> hmm
2015-06-02 07:45:34 <ncopa> but you dont need to bootstrap gcc again
2015-06-02 07:45:45 <fabled> no piepatches for gcc-5.1 yet
2015-06-02 07:45:55 <ncopa> once isl is updated, gcc will work but without isl
2015-06-02 07:46:02 <ncopa> til gcc itself is rebuild
2015-06-02 07:46:30 <fabled> sounds like that
2015-06-02 07:46:33 <ncopa> i think it might make sense to add that patch
2015-06-02 07:46:53 <ncopa> otherwise we might need bootstrap gcc/isl to upgrade isl
2015-06-02 07:55:20 <fabled> ncopa, what does gcc-pure64.patch do ?
2015-06-02 07:56:02 <fabled> sounds like it can go
2015-06-02 07:56:10 <ncopa> probably
2015-06-02 07:56:19 <ncopa> i think it might be to remove multilib
2015-06-02 07:56:26 <ncopa> eg. not install things in /usr/lib64
2015-06-02 07:56:52 <fabled> oh, it could b
2015-06-02 07:57:13 <ncopa> instead, for 64 bit look in /usr/lib/ and for 32bit (multilib) look in /usr/lib32
2015-06-02 07:57:55 <ncopa> reminds me
2015-06-02 07:58:05 <ncopa> i really want fix -m32 with musl
2015-06-02 07:58:33 <ncopa> in glibc then headers are both 32 and 64 bit mixed in same headers
2015-06-02 07:58:53 <ncopa> so if you do gcc -m32 with glibc, it will pick the right sizes
2015-06-02 07:59:08 <ncopa> but with musl, they have completely separate headers for 32bit and 64bit
2015-06-02 07:59:15 <ncopa> so with gcc -m32
2015-06-02 07:59:20 <ncopa> you get wrong sizes
2015-06-02 07:59:38 <ncopa> that was an issue with xen hvmloader which is complied with -m32
2015-06-02 07:59:40 <fabled> so 64-bit stuff should also do 32-bit headers, and patch gcc to use them with -m32 ?
2015-06-02 07:59:56 <fabled> hmm
2015-06-02 08:00:12 <ncopa> i think the proposed solution from dalias was
2015-06-02 08:00:36 <ncopa> on 64 bit, also install the 32 bit headers, but in other location
2015-06-02 08:01:13 <ncopa> then with -m32, inject -I/usr/include32
2015-06-02 08:01:19 <ncopa> or wherever the inludes ends up
2015-06-02 08:02:22 <ncopa> i think it was stdint.h atleast
2015-06-02 08:03:00 <ncopa> http://git.alpinelinux.org/cgit/aports/tree/main/xen/musl-hvmloader-fix-stdint.patch
2015-06-02 08:03:57 <ncopa> together with this: http://git.alpinelinux.org/cgit/aports/tree/main/xen/stdint_local.h
2015-06-02 08:04:00 <ncopa> actually
2015-06-02 08:04:10 <ncopa> i think it might be worth do that in musl 64 bit headers...
2015-06-02 08:04:57 <ncopa> looks like it is only int64_t and uint64_t
2015-06-02 08:07:45 <fabled> i wonder if we should take http://port70.net/~nsz/musl/.gcc/libgcc.diff
2015-06-02 08:08:30 <ncopa> what does that do?
2015-06-02 08:08:37 <ncopa> disable symver for libgcc?
2015-06-02 08:10:32 <ncopa> does musl provide a __cpu_indicator_init?
2015-06-02 08:10:34 <ncopa> or glibc?
2015-06-02 08:11:42 <ncopa> looks like this: https://gcc.gnu.org/ml/gcc-bugs/2015-05/msg00048.html
2015-06-02 08:18:39 <fabled> ncopa, no i think it was some issue with musl; there was thread about it in the gcc mailing list
2015-06-02 08:19:02 <fabled> https://gcc.gnu.org/ml/gcc-patches/2015-05/msg00829.html
2015-06-02 08:20:13 <fabled> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65612
2015-06-02 08:21:46 <fabled> iirc, that did not work with musl due to some strange use of versioned symbols
2015-06-02 08:22:25 <fabled> should probably take it
2015-06-02 08:22:50 <ncopa> yeah
2015-06-02 08:27:01 <ncopa> rnalrd: thanks for fixing mini_httpd
2015-06-02 08:27:15 <rnalrd> yw
2015-06-02 08:28:43 <fabled> ok, i have other patches rebased for gcc 5.1 now except piepatches
2015-06-02 08:57:43 <fabled> patches done, doing first test build
2015-06-02 09:09:45 <nsz> fabled: if you move to gcc 6 (or possibly gcc 5.2) then musl will not be able to load libgcc_s.so.1
2015-06-02 09:10:33 <fabled> so it's not affecting gcc 5.1 ?
2015-06-02 09:10:46 <nsz> 5.1 does not have the issue yet (there was a bug in x86 multiversioning feature and the versioning workaround in gcc trunk does not work on musl)
2015-06-02 09:11:09 <fabled> i tired applying that libgcc patch, but it fails
2015-06-02 09:11:17 <fabled> apparently due to that then
2015-06-02 09:11:32 <nsz> yes it needs the symver workaround from gcc 6 to be backported
2015-06-02 09:11:47 <fabled> oh?
2015-06-02 09:11:56 <fabled> is it planned upstream?
2015-06-02 09:12:40 <nsz> i'm sure they will eventually back port it to branch 5 since without that the multiversioning feature is broken for pie and dso
2015-06-02 09:13:33 <nsz> so most likely in 5.2 or so  there will be a versioned symbol in libgcc_s.so without a default version
2015-06-02 09:14:00 <nsz> (the symbol is referenced from libgcc_s.so because it's a constructor function)
2015-06-02 09:14:19 <nsz> and that is an undefined symbol as far as musl is concerned (currently)
2015-06-02 09:14:53 <nsz> so when that happens, you'll just have to remove that symbol (or apply my patch)
2015-06-02 09:15:23 <fabled> ok
2015-06-02 09:15:27 <nsz> but it's also possible that by that time musl will have a workaround
2015-06-02 09:15:33 <fabled> i'll try building now 5.1.0
2015-06-02 09:15:39 <fabled> with all alpine patches in
2015-06-02 09:15:49 <fabled> and musl patches from your stash
2015-06-02 09:15:54 <fabled> so let's see how that goes
2015-06-02 09:32:17 <ncopa> fabled: i wonder how to deal with symlink depends
2015-06-02 09:33:31 <ncopa> i wonder if we should introduce "provides = symlinktarget:<file>"
2015-06-02 09:33:52 <ncopa> and let apk figure out which package provides the target of symlink
2015-06-02 09:34:31 <ncopa> the idea is that abuild scans the staging dir for symlinks
2015-06-02 09:34:38 <ncopa> find $dir -type l
2015-06-02 09:35:08 <ncopa> and for each broken symlink it adds it to a list which it will need to figure out the depends
2015-06-02 09:35:49 <ncopa> the intention is to fix -dev subpackage to only pull in what they need
2015-06-02 09:35:55 <ncopa> and not everything like we currently do
2015-06-02 09:38:29 <fabled> dunno
2015-06-02 09:38:35 <ncopa> i suppose that is a bad idea
2015-06-02 09:39:03 <ncopa> because unlike .so files, symlink targets (regular file) has nothing that says it is a symlink target
2015-06-02 09:40:22 <fabled> nsz, hum, did you have patch for all that: error: missing binary operator before token "("
2015-06-02 09:40:22 <fabled>  #if __GLIBC_PREREQ(2,15) && defined(_GNU_SOURCE)
2015-06-02 09:40:22 <fabled>                    ^
2015-06-02 09:40:22 <fabled> "
2015-06-02 09:42:55 <nsz> where is this?
2015-06-02 09:43:10 <nsz> libitm had some patch like this
2015-06-02 09:45:14 <nsz> (i didnt build fortran/java/ada so some extra patching might be needed there)
2015-06-02 09:58:19 <fabled> libstdc++-v3/include/x86_64-alpine-linux-musl/bits/os_defines.h:44:19:
2015-06-02 09:58:19 <fabled> error: missing binary operator before token "("
2015-06-02 09:58:26 <fabled> gcc-5.1.0
2015-06-02 10:00:20 <fabled> ./src/build/x86_64-alpine-linux-musl/libstdc++-v3/include/x86_64-alpine-linux-musl/bits/os_defines.h -> /home/tteras/aports/main/gcc/src/gcc-5.1.0/libstdc++-v3/config/os/gnu-linux/os_defines.h
2015-06-02 10:00:43 <nsz> you are supposed to use os/generic/os_defines.h
2015-06-02 10:01:21 <fabled> hu
2015-06-02 10:01:24 <fabled> hum
2015-06-02 10:01:27 <nsz> i have that in my gthr patch
2015-06-02 10:02:52 <fabled> strange
2015-06-02 10:02:53 <fabled> it's missing
2015-06-02 10:03:00 <fabled> i messed up by build somehow
2015-06-02 10:06:39 <fabled> i'll take all those i need individually then
2015-06-02 11:08:57 <fabled>         so:libz.so.1
2015-06-02 11:08:57 <fabled> >>> g++*: Package size: 61.5 MB
2015-06-02 11:08:57 <fabled> >>> g++*: Compressing data...
2015-06-02 11:08:57 <fabled> tar: usr/include/c++/5.1.0/gnu/awt: Read error at byte 0, while reading 4096 bytes: Is a directory
2015-06-02 11:08:57 <fabled> tar: Exiting with failure status due to previous errors
2015-06-02 11:09:05 <fabled> huh
2015-06-02 11:09:12 <fabled> strange error
2015-06-02 11:15:16 <nsz> dmesg?
2015-06-02 11:15:39 <nsz> (can be a hw issue if tar cannot read something)
2015-06-02 11:16:21 <nsz> hm maybe not if it was a directory
2015-06-02 11:16:31 <fabled> nothing in dmesg
2015-06-02 11:17:22 <fabled> >>> g++*: Scanning shared objects
2015-06-02 11:17:22 <fabled> scanelf: mmap on '/home/tteras/aports/main/gcc/pkg/g++/usr/include/c++/5.1.0/gnu/java/rmi' of 4096 bytes failed :(
2015-06-02 11:17:28 <fabled> somehow it thinks it's a file
2015-06-02 11:19:08 <fabled> must be fakeroot
2015-06-02 11:19:20 <fabled> it works without fakeroot
2015-06-02 11:24:24 <fabled> $ gcc hello.c  -o hello
2015-06-02 11:24:25 <fabled> ttdev-edge-x86_64:~$ ./hello
2015-06-02 11:24:25 <fabled> Hello gcc 5.1!
2015-06-02 11:24:28 <fabled> $ gcc --version
2015-06-02 11:24:28 <fabled> gcc (Alpine 5.1.0) 5.1.0
2015-06-02 11:24:57 <fabled> at least it's not completely broken
2015-06-02 11:25:18 <fabled> it's doing PIE, relro is there too
2015-06-02 11:25:28 <fabled> and bind now
2015-06-02 11:27:41 <fabled> hmm
2015-06-02 11:27:44 <fabled> java specs are in wrong place
2015-06-02 11:45:29 <fabled> humm
2015-06-02 11:45:45 <fabled> so gcc-5.1 looks otherwise ok, except libgcj.spec is mislocated / not found by gcj
2015-06-02 11:45:47 <fabled> dunno why
2015-06-02 11:46:22 <fabled> tested only on x86_64, though
2015-06-02 11:47:33 <nsz> is it a native gcc build? or cross compiled?
2015-06-02 11:48:05 <fabled> should be native, but it might be gcj thinking to be a cross-compiler
2015-06-02 11:48:20 <fabled> since the last spec check is appending the toolchain triplet
2015-06-02 11:48:31 <nsz> if native then you may want to run 'make check' ..only takes a few hours :)
2015-06-02 11:48:31 <fabled> which is not done in other frontends (fortran/ada/gcc)
2015-06-02 11:50:45 <nsz> ..and then run contrib/test_summary which should have similar output to http://port70.net/~nsz/musl/.gcc/test_summary.txt
2015-06-02 11:52:16 <fabled> it needs some additional stuff?
2015-06-02 11:52:20 <fabled> autogen -T /home/tteras/aports/main/gcc/src/gcc-5.1.0/fixincludes/check.tpl /home/tteras/aports/main/gcc/src/gcc-5.1.0/fixincludes/inclhack.def
2015-06-02 11:52:20 <fabled> /bin/bash: autogen: command not found
2015-06-02 11:53:03 <nsz> hm yes, most likely it needs dejagnu, tcl and who knows what else
2015-06-02 11:53:25 <fabled> ncopa, maybe i just move usr/lib/*.spec to the usr/lib/gcc/x86_64-alpine-linux-musl/5.1.0/
2015-06-02 11:56:21 <ncopa> ok
2015-06-02 11:56:56 <nsz> fabled: i think you probably dont need that autogen, if you run 'make -k check' it will keep going
2015-06-02 11:57:04 <nsz> but dejagnu, tcl, expect is needed
2015-06-02 11:57:59 <nsz> (autogen is only for fixincludes, which we dont care about)
2015-06-02 12:11:21 <fabled> what was that libstdc++ abi change in gcc5.1
2015-06-02 12:11:33 <fabled> https://gcc.gnu.org/onlinedocs/libstdc++/manual/using_dual_abi.html
2015-06-02 12:54:00 <nsz> i think the new c++11 symbols have different mangled names because of the new c++ inline namespace, so you should get the old and new abi too
2015-06-02 13:00:38 <nsz> i guess there are problems if std::string or std::list is passed across dso boundaries, where dsos are built with different abi
2015-06-02 13:04:25 <fabled> well, the docs say the .so ships both
2015-06-02 13:04:34 <fabled> so it should be backwards compatible
2015-06-02 13:04:54 <fabled> unless their code does not support mixing of the namespace usages
2015-06-02 13:11:22 <nsz> void f(std::string s){} will have different type depending on the value of _GLIBCXX_USE_CXX11_ABI at compile time
2015-06-02 13:11:56 <nsz> so even though libstdc++so has dual abi, the dsos you build are not dual abi but either new or old
2015-06-02 13:13:01 <nsz> by defaul they will have new abi, but if there are references across dsos with different abi then there will be missing symbols
2015-06-02 13:13:33 <nsz> and then you will have to downgrade your code to old abi with -D_GLIBCXX_USE_CXX11_ABI=0
2015-06-02 13:14:46 <nsz> s/will have different type/will have different mangled name/
2015-06-02 13:45:14 <fabled> oh
2015-06-02 13:45:15 <fabled> right
2015-06-02 14:31:23 <clandmeter> ncopa: i cannot call chown from apkbuild anymore?
2015-06-02 14:32:23 <ncopa> clandmeter: what error do you get?
2015-06-02 14:32:27 <ncopa> does the user exist?
2015-06-02 14:32:38 <ncopa> do you have permissions to do chown?
2015-06-02 14:34:07 <clandmeter> ncopa: trying to bump http://git.alpinelinux.org/cgit/aports/tree/main/oscam/APKBUILD
2015-06-02 14:34:26 <clandmeter> chown: /home/clandmeter/aports/main/oscam/pkg/oscam/etc/oscam: Operation not permitted
2015-06-02 14:35:23 <clandmeter> which permissions should i have? i didnt change anuthing
2015-06-02 14:35:26 <clandmeter> anything
2015-06-02 14:36:54 <ncopa> what permissions is /home/clandmeter/aports/main/oscam/pkg/oscam/etc set to?
2015-06-02 14:37:46 <ncopa> 	install -D -m644 "$srcdir"/$pkgname.conf \
2015-06-02 14:37:46 <ncopa> 		"$pkgdir"/etc/$pkgname/$pkgname.conf || return 1
2015-06-02 14:37:54 <ncopa> it must be that one that fails
2015-06-02 14:39:06 <clandmeter> ncopa: you think? isnt it line 71?
2015-06-02 14:39:20 <ncopa> ah
2015-06-02 14:39:22 <ncopa> yes
2015-06-02 14:39:24 <ncopa> probably
2015-06-02 14:39:38 <clandmeter> i just dont know why
2015-06-02 14:40:07 <ncopa> could be a problem in recent fakeroot
2015-06-02 14:43:25 <clandmeter> any workaround for now?
2015-06-02 14:44:35 <ncopa> i dont know
2015-06-02 14:44:42 <ncopa> i dont know why it happens either
2015-06-02 14:45:32 <ttilley> fabled: re: 5.1, one of the musl team's patches didn't make it upstream, but it's present in nsz's patchset. he has it in a dotdir on his host so i'll send it to you in a msg
2015-06-02 14:45:44 <ncopa> i can look at it after i'm done with a haserl issue and after i'm done with abuild's -dev fixes and after I have looked at Scrumpyjacks patches (and i should probably look at the other ppls patches too....)
2015-06-02 14:45:50 <fabled> ttilley, i have gcc-5.1 commit ready
2015-06-02 14:46:07 <fabled> ttilley, http://sprunge.us/PTCb
2015-06-02 14:46:10 <ttilley> lol, ok
2015-06-02 14:46:19 <ttilley> sorry, just got to the computer :)
2015-06-02 14:46:27 <fabled> it builds, and i've verified at to some extent
2015-06-02 14:46:29 <ncopa> clandmeter: if you find out that it is a problem in abuild itself, then please let me know. i am supposed to work on abuild now anyway
2015-06-02 14:46:35 <ttilley> awesome
2015-06-02 14:51:15 <ncopa> clandmeter: nasty. i tried building unmodified oscam
2015-06-02 14:51:20 <ncopa> it built, not problem
2015-06-02 14:51:22 <ncopa> i did apk upgrade
2015-06-02 14:51:23 <ncopa> and rebuild
2015-06-02 14:51:25 <ncopa> and it fails
2015-06-02 14:52:13 <ncopa> from what i can see, nothing relevant fakeroot has changed
2015-06-02 14:52:42 <ahills> it fails for me outside of fakeroot...
2015-06-02 14:52:54 <ahills> I just export srcdir and pkgdir, then . APKBUILD, then "package"
2015-06-02 14:54:51 <clandmeter> weird
2015-06-02 14:55:12 <ncopa> yes
2015-06-02 14:55:15 <ahills> I also can't su in this container
2015-06-02 14:55:24 <clandmeter> something in musl changed?
2015-06-02 14:55:25 <ahills> bbsuid is ---s--x--x, is that correct?
2015-06-02 14:55:40 <ncopa> there is a recent busybox change
2015-06-02 14:55:51 <ahills> yes, after updating busybox I can't su
2015-06-02 14:55:55 <ahills> su: can't set groups: Operation not permitted
2015-06-02 14:55:57 <ncopa> which sets capabilities
2015-06-02 14:56:17 <ahills> sudo works fine
2015-06-02 14:56:23 <ncopa> might be relevant
2015-06-02 14:57:16 <ahills> will try downgrading busybox
2015-06-02 14:57:48 <ahills> yeah, 1.23.2-r0->r1 broke for me
2015-06-02 14:57:55 <ahills> broke bbsuid I guess
2015-06-02 15:01:38 <ahills> does someone have a quick link to a cap reference? digging through http://www.friedhoff.org/posixfilecaps.html
2015-06-02 15:02:02 <ncopa> im not convienced that is theproblem
2015-06-02 15:02:34 <ahills> I'm not either until I read what the cap setcap cap_net_raw+ep "$pkgdir"/bin/busybox means
2015-06-02 15:03:30 <ahills> but it *is* nice to have su working on my machine again ;)
2015-06-02 15:06:38 <ncopa> ahills: ok, im sure
2015-06-02 15:06:51 <ncopa> setcap broke su for you
2015-06-02 15:07:07 <ncopa> setcap -r /bin/busybox makes it work again
2015-06-02 15:07:41 <ncopa> kaniini_: do you mind have a look at why setcap breaks su and chown in busybox?
2015-06-02 15:07:43 <ahills> bbsuid just execv's busybox right?
2015-06-02 15:07:48 <ncopa> yes
2015-06-02 15:07:53 <ncopa> i think what happens
2015-06-02 15:08:04 <ncopa> when capabilities extended attribute is set
2015-06-02 15:08:20 <ncopa> it will remove all those capabilites that are not explicitly set
2015-06-02 15:08:22 <ncopa> even from root
2015-06-02 15:08:52 <ncopa> so busybox chown and su breaks, even if it is root
2015-06-02 15:09:15 <ncopa> temp workaround is: setcap -r /bin/busybox
2015-06-02 15:09:25 <ncopa> (but that will break ping)
2015-06-02 15:09:35 <ncopa> argH!
2015-06-02 15:09:37 <ncopa> time is up
2015-06-02 15:09:47 <ncopa> and i didnt fix haserl issue
2015-06-02 15:35:16 <ahills> time is up??
2015-06-02 15:35:51 <ahills> I've just pinned busybox at -r0
2015-06-02 15:35:58 <ahills> so my ping, etc still work
2015-06-02 15:37:53 <ahills> clandmeter / ncopa: building oscam succeeds once I revert busybox to 1.23.2-r0
2015-06-02 15:53:27 <kaniini_> ncopa: hmm
2015-06-02 15:53:38 <kaniini_> ncopa: perhaps we just give the permissions to bbsuid instead
2015-06-02 15:59:51 <ahills> it is the basic limitation of having a single executable...
2015-06-02 16:47:29 <ScrumpyJack> no tteras today?
2015-06-02 16:47:38 <fabled> ScrumpyJack, ?
2015-06-02 16:47:46 <fabled> i'm here. though going home in a moment
2015-06-02 16:48:52 <ScrumpyJack> ah ok. you contributed libreoffice right?
2015-06-02 16:50:05 <ScrumpyJack> we're trying to make the Mate desktop look good for setup-box and one of the (little annoying) things is that the libreoffice icons don't show up in the mate-menu
2015-06-02 16:50:29 <ScrumpyJack> is this something you could help us with?
2015-06-02 16:54:43 <fabled> yes, i made the libreoffice apkbuild
2015-06-02 16:54:58 <fabled> sounds just like wrong icon names, or the iconset not being installed
2015-06-02 16:55:05 <fabled> maybe email on alpine-devel?
2015-06-02 16:55:07 <fabled> i need to run
2015-06-02 17:28:38 <alacerda> http://pastebin.com/5ERn2feq
2015-06-02 17:28:43 <alacerda> ScrumpyJack, ^^
2015-06-02 17:28:55 <ScrumpyJack> that is from apk update && apk upgrade && apk  search llvm
2015-06-02 17:29:03 <ScrumpyJack> in edge
2015-06-02 22:30:33 <BitL0G1c> trying to build a package with xorg-server-dev gives ' so:libLLVM-3.6.so (missing):' http://hastebin.com/ixacoliqig.avrasm
2015-06-02 22:31:38 <BitL0G1c> this is in edge with apk updated & upgraded
2015-06-03 00:08:32 <ahills> Will anyone care if I break the redirect to Arch Linux dm-crypt instructions for setting up encrypted systems on the wiki? There are some notable differences, e.g. mkinitfs vs. mkinitcpio, kernel parameters, etc
2015-06-03 02:46:03 <ahills> ncopa: are you interested in patches for alpine-conf that add cryptsetup integration to setup-disk?
2015-06-03 04:26:24 <vitronic> ahills, public patch
2015-06-03 05:52:46 <crow> ahills i think BitL0G1c did already something in that way http://it-offshore.co.uk/linux/alpine-linux/25-alpine-linux-luks-encrypted-installations
2015-06-03 06:00:15 <ncopa> morning
2015-06-03 06:01:00 <ncopa> kaniini_: it would be nice if you could investigate the su issue and the chown in fakeroot (build main/oscam)
2015-06-03 06:01:04 <ncopa> currently things are broken
2015-06-03 06:01:56 <ncopa> BitL0G1c: what arch is it? x86? x86_64? if you do 'apk version', do you see any package not beeing upgraded?
2015-06-03 06:02:09 <ncopa> i think the 'apk upgrade -U -a' didnt upgrade mesa
2015-06-03 06:02:13 <ncopa> so you get the error
2015-06-03 06:03:42 <ncopa> ahills: yes, i think it would be nice with cryptsetup in setup-disk
2015-06-03 06:03:53 <ncopa> i'd like rewrite the setup-disk tool from scratch though...
2015-06-03 06:10:10 <ncopa> we need uefi support etc
2015-06-03 06:34:20 <ncopa> fabled: what do you think about adding provides="mysql-dev" in mariadb-dev? http://bugs.alpinelinux.org/issues/4264
2015-06-03 06:35:20 <fabled> ncopa, should be ok
2015-06-03 06:35:25 <fabled> as long as they are interchangeable
2015-06-03 06:35:45 <ncopa> mysql-dev no longer exist
2015-06-03 06:36:02 <ncopa> what i did was sed 's/mysql-dev/mariadb-dev/g'
2015-06-03 06:36:05 <fabled> you can do provides="mysql-dev=$pkgver-r$pkgrel" if you want to add preference
2015-06-03 06:36:06 <ncopa> in all APKBUILDs
2015-06-03 06:36:19 <ncopa> ok
2015-06-03 06:36:20 <ncopa> yeah
2015-06-03 06:36:22 <ncopa> makes sense
2015-06-03 06:40:45 <ncopa> it would hide the problem that mysql-dev is used when mariadv-dev should been used
2015-06-03 06:40:52 <ncopa> i dont know what we prefer
2015-06-03 06:40:59 <ncopa> i suppose it is a nice thing to not break things
2015-06-03 06:41:06 <ncopa> be backwards compat whenever possible
2015-06-03 06:51:07 <ncopa> i wonder if i should add provides for pkg-config files
2015-06-03 06:51:19 <ncopa> eg. provides=pc:openssl
2015-06-03 06:51:44 <ncopa> then packages who needs pkg-config openssl can use: depend="pc:openssl"
2015-06-03 06:51:54 <ncopa> similar to what we do with so:* depends
2015-06-03 06:52:48 <ncopa> i plan to add supoprt for abuild find the pkg-config dependencies
2015-06-03 06:53:04 <ncopa> $ pkg-config --print-requires openssl
2015-06-03 06:53:04 <ncopa> libssl
2015-06-03 06:53:04 <ncopa> libcrypto
2015-06-03 06:54:05 <ncopa> so openssl-dev package (or whatever package that provides openssl.pc) would get: depends="pc:libssl pc:libcrypto"
2015-06-03 06:54:17 <ncopa> that way we could track the pk
2015-06-03 06:54:24 <ncopa> that way we could track the pkg-config dependencies
2015-06-03 07:04:00 <kaniini_> ncopa: i'm reverting it for now
2015-06-03 07:04:06 <ncopa> ok
2015-06-03 07:04:11 <kaniini_> ncopa: i spent yesterday looking at it and couldn't find a good answer
2015-06-03 07:04:14 <ncopa> sounds good
2015-06-03 07:04:26 <ncopa> yes, i got the feeing it was non-trivial so solve
2015-06-03 07:04:51 <ncopa> the linux caps are difficult to actually use...
2015-06-03 07:05:32 <kaniini_> instead i think what we do is accept the suid binary situation for busybox
2015-06-03 07:05:51 <ncopa> we could probably harden bbsuid a bit though
2015-06-03 07:05:53 <kaniini_> and bring it up as an issue against toybox
2015-06-03 07:05:56 <ncopa> clean env
2015-06-03 07:06:01 <kaniini_> so that maybe toybox will fix it up
2015-06-03 07:06:02 <kaniini_> ;)
2015-06-03 07:06:11 <ncopa> probably better yes
2015-06-03 07:06:30 <ncopa> kaniini_: while i have you here
2015-06-03 07:06:32 <kaniini_> TCB shadow would be nice to have
2015-06-03 07:06:53 <ncopa> im working on abuild autoamtically detect dependencies for pkg-config files
2015-06-03 07:07:23 <ncopa> i want list all deps of $pkgdir/usr/lib/pkgocnfig/*.pc files
2015-06-03 07:07:49 <kaniini_> well, first thing first you'd have a provides=pkg-config:<modulename> probably
2015-06-03 07:08:00 <kaniini_> do you want some stuff to just extract parameters from a pkg-config file?
2015-06-03 07:08:09 <ncopa> i suppose PKG_CONFIG_PATH="$pkgdir/usr/lib/pkgconfig" env var should do it
2015-06-03 07:08:15 <kaniini_> yes
2015-06-03 07:08:32 <kaniini_> --list-all will give you all module names it finds
2015-06-03 07:08:46 <ncopa> aha
2015-06-03 07:08:52 <kaniini_> i could add a --list-module-names which just gives the module names
2015-06-03 07:09:17 <ncopa> PKG_CONFIG_PATH="$pkgdir/usr/lib/pkgconfig" pkg-config --list-module-names
2015-06-03 07:09:22 <ncopa> hum
2015-06-03 07:09:30 <ncopa> that also measn that abuild will depend on pkg-config
2015-06-03 07:09:56 <ncopa> i suppose thats not a big deal
2015-06-03 07:10:11 <ncopa> alternatively i use awk for it
2015-06-03 07:10:14 <kaniini_> you could make it an install_if
2015-06-03 07:10:21 <kaniini_> to break the cycle
2015-06-03 07:10:26 <ncopa> nah
2015-06-03 07:10:38 <ScrumpyJack> morning
2015-06-03 07:10:41 <kaniini_> i would not advise running awk against a .pc file
2015-06-03 07:10:43 <ncopa> i suppose pkgconf is easy to crosscompile?
2015-06-03 07:10:45 <kaniini_> yes
2015-06-03 07:10:52 <kaniini_> pkgconf is ISO C90
2015-06-03 07:11:02 <kaniini_> (well, with stdbool)
2015-06-03 07:11:08 <ncopa> better add it as dependency for abuild
2015-06-03 07:11:22 <kaniini_> i think build-base pulls it in anyway
2015-06-03 07:11:33 <ncopa> i think not
2015-06-03 07:11:39 <kaniini_> hmm
2015-06-03 07:11:41 <ncopa> but most -dev packages pulls it it
2015-06-03 07:11:56 <ncopa> might be that build-base depends on a -dev package that depends on pkg-config
2015-06-03 07:12:16 <kaniini_> i'll add a --list-module-names
2015-06-03 07:13:39 <ncopa> hum
2015-06-03 07:14:07 <ncopa> i think i need it to exclude sysmten pc modules
2015-06-03 07:14:25 <ncopa> for the provides part
2015-06-03 07:15:48 <kaniini_> if you restrict PKG_CONFIG_PATH it should
2015-06-03 07:15:54 <ncopa> or i can simply do: for i in usr/lib/pkgconfig/*.pc; module=$(basename $i .pc); provides="pc:$module"; done
2015-06-03 07:16:41 <ncopa> ncdev-edge-musl-x86:~/abuild$ PKG_CONFIG_PATH=/tmp/foo/bar pkg-config --list-all
2015-06-03 07:16:41 <ncopa> openssl                        OpenSSL - Secure Sockets Layer and cryptography libraries and tools
2015-06-03 07:16:41 <ncopa> libssl                         OpenSSL-libssl - Secure Sockets Layer and cryptography libraries
2015-06-03 07:16:41 <ncopa> libcrypto                      OpenSSL-libcrypto - OpenSSL cryptography library
2015-06-03 07:16:41 <ncopa> zlib                           zlib - zlib compression library
2015-06-03 07:16:42 <ncopa> libgcj-4.9                     libgcj - libgcj
2015-06-03 07:16:48 <ncopa> should been empty?
2015-06-03 07:19:27 <ncopa> http://sprunge.us/caWM
2015-06-03 07:19:41 <fabled> on upgrade i got:
2015-06-03 07:19:42 <fabled> (3/3) Upgrading busybox-static (1.23.2-r1 -> 1.23.2-r2)
2015-06-03 07:19:42 <fabled> Executing busybox-1.23.2-r2.trigger
2015-06-03 07:19:42 <fabled> var/cache/misc/busybox-1.23.2-r2.trigger: line 19: /bin/bbsuid: not found
2015-06-03 07:20:00 <ncopa> is busybox installed?
2015-06-03 07:20:10 <fabled> yes
2015-06-03 07:20:11 <ncopa> hum
2015-06-03 07:20:13 <ncopa> yes
2015-06-03 07:20:18 <ncopa> its busybox trigger
2015-06-03 07:20:30 <fabled> sorry
2015-06-03 07:20:31 <fabled> (1/3) Upgrading busybox (1.23.2-r1 -> 1.23.2-r2)
2015-06-03 07:20:31 <fabled> Executing busybox-1.23.2-r2.post-upgrade
2015-06-03 07:20:31 <fabled> (2/3) Upgrading acf-lib (0.7.3-r0 -> 0.7.3-r1)
2015-06-03 07:20:31 <fabled> (3/3) Upgrading busybox-static (1.23.2-r1 -> 1.23.2-r2)
2015-06-03 07:20:33 <fabled> Executing busybox-1.23.2-r2.trigger
2015-06-03 07:20:35 <fabled> var/cache/misc/busybox-1.23.2-r2.trigger: line 19: /bin/bbsuid: not found
2015-06-03 07:21:20 <ncopa> sounds like bb is broke...
2015-06-03 07:24:08 <fabled> bbsuid is not in the package
2015-06-03 07:24:11 <kaniini_> wtf.
2015-06-03 07:24:19 <kaniini_> all i did was git revert
2015-06-03 07:24:25 <fabled> not sure if it's apkbuild error, or fakeroot/tar issue
2015-06-03 07:24:29 <kaniini_> and then bump the pkgrel
2015-06-03 07:25:19 <ncopa> http://bld2.alpinelinux.org/buildlogs/build-edge-x86_64/main/busybox/busybox-1.23.2-r2.log
2015-06-03 07:25:25 <ncopa> tar: bin/bbsuid: Cannot open: Permission denied
2015-06-03 07:25:25 <ncopa> tar: Exiting with failure status due to previous errors
2015-06-03 07:25:40 <fabled> huh
2015-06-03 07:25:54 <ncopa> previous setcap'ed busybox was broke
2015-06-03 07:26:01 <ncopa> so kaniini_ reverted it
2015-06-03 07:26:14 <fabled> builder should have GNU tar
2015-06-03 07:26:21 <fabled> i wonder
2015-06-03 07:26:29 <fabled> if GNU tar checks permissions even when running as root
2015-06-03 07:26:33 <fabled> or if it's fakeroot issue
2015-06-03 07:26:43 <fabled> it's -m4111
2015-06-03 07:26:46 <ncopa> smells fakeroot
2015-06-03 07:26:47 <fabled> which is execute all
2015-06-03 07:26:48 <ncopa> however
2015-06-03 07:26:56 <ncopa> bb was setcap'ed
2015-06-03 07:27:03 <ncopa> so the *shell* was setcaped
2015-06-03 07:27:03 <kaniini_> i think it should be -m4755
2015-06-03 07:27:05 <fabled> should've been using GNU tar
2015-06-03 07:27:22 <ncopa> hum
2015-06-03 07:27:40 <ncopa> could also been fakeroot isses?
2015-06-03 07:27:46 <fabled> yes
2015-06-03 07:27:47 <ncopa> but we havent touched fakeroot recently
2015-06-03 07:27:56 <fabled> i saw some strange issue also when building gcc
2015-06-03 07:28:02 <fabled> it thought a directory is a file
2015-06-03 07:28:15 <kaniini_> i wonder if that is related to the bootstrap issue i am facing on CI20
2015-06-03 07:28:18 <ncopa> oscam build broke on chown
2015-06-03 07:28:28 <fabled> i wonder if there's some musl breakage
2015-06-03 07:28:52 <ncopa> hum
2015-06-03 07:28:53 <kaniini> i think when fabled bootstrapped arm/musl he used a uclibc machine to do it
2015-06-03 07:29:04 <kaniini> because we weren't building x86/musl yet
2015-06-03 07:29:05 <ncopa> i think this is bb issue
2015-06-03 07:29:20 <fabled> it could bb/musl/gnu-tar/fakeroot
2015-06-03 07:29:34 <ncopa> what i know for sure
2015-06-03 07:29:50 <ncopa> oscam failed with error yesterday on chown (inside fakeroot)
2015-06-03 07:30:00 <ncopa> setcap -r /bin/busybox
2015-06-03 07:30:05 <ncopa> and it works again
2015-06-03 07:30:21 <kaniini> hmm
2015-06-03 07:30:49 <ncopa> >>> busybox: Building bbsuid
2015-06-03 07:32:10 <ncopa>         install -m4111 bbsuid "$pkgdir"/bin/bbsuid || return 1
2015-06-03 07:32:23 <ncopa> that is inside fakeroot
2015-06-03 07:34:34 <ncopa> yeah
2015-06-03 07:34:47 <ncopa> i manually setcap on my dev box
2015-06-03 07:34:53 <ncopa> rebuilding busybox worked fine here
2015-06-03 07:35:04 <ncopa> i think I'll need to manually fix the edge builders
2015-06-03 07:36:09 <kaniini> so sorry about that
2015-06-03 07:36:16 <ncopa> np
2015-06-03 07:36:17 <ncopa> happens
2015-06-03 07:36:52 <kaniini> so what will you use the pkg-config deps for
2015-06-03 07:36:52 <ncopa> without risk breaking anything you will never be able to make progress :)
2015-06-03 07:37:13 <ncopa> automatically add depepends to -dev packages
2015-06-03 07:37:23 <ncopa> if a -dev package ships a .pc file
2015-06-03 07:37:34 <ncopa> abuild should figure out the depends for it
2015-06-03 07:37:48 <ncopa> currently you will have to do that manually (thats why depends_dev exist basically)
2015-06-03 07:39:18 <kaniini> that only works if we have a .pc file though :(
2015-06-03 07:39:35 <kaniini> i guess you also want something to parse the .pc file and dump its dependency tree?
2015-06-03 07:40:00 <kaniini> bapt@freebsd was thinking about that few months ago
2015-06-03 07:40:13 <ncopa> nah, not yet
2015-06-03 07:40:17 <ncopa> i dont need that
2015-06-03 07:40:26 <ncopa> i think currently i only wat abuild automatically add:
2015-06-03 07:40:37 <ncopa> provides=pc:<module>
2015-06-03 07:40:43 <ncopa> and depends=pc:<module>
2015-06-03 07:41:29 <ncopa> the depends=pc:<module> should not be added if same package also has provides for same module
2015-06-03 07:41:40 <fabled> ncopa, why setcap -r was needed?
2015-06-03 07:42:02 <fabled> apk should've removed it
2015-06-03 07:42:26 <fabled> hmm
2015-06-03 07:42:29 <ncopa> it did
2015-06-03 07:42:43 <ncopa> i think just rebuilding busybox shoudl do it
2015-06-03 07:42:52 <ncopa> after second run it worked
2015-06-03 07:42:57 <fabled> oh
2015-06-03 07:43:05 <ncopa> so fix is simple
2015-06-03 07:43:14 <ncopa> just remove the previous build
2015-06-03 07:43:19 <ncopa> and trigger rebuild
2015-06-03 07:43:35 <ncopa> trigger new upload that is
2015-06-03 07:45:57 <ncopa> fabled: try apk upgrade -U -a now
2015-06-03 07:47:44 <fabled> yeah, it's good now
2015-06-03 07:47:59 <ncopa> ok
2015-06-03 07:48:04 <ncopa> what is worrying though is
2015-06-03 07:48:08 <ncopa> why didnt it abort
2015-06-03 07:48:19 <ncopa> instead of build a broken package
2015-06-03 07:48:27 <ncopa> it should have exited with error
2015-06-03 07:58:05 <AmatCoder> ncopa: same issue than BitL0G1c here...
2015-06-03 07:58:11 <AmatCoder> ERROR: unsatisfiable constraints:
2015-06-03 07:58:16 <AmatCoder>   so:libLLVM-3.6.so (missing):
2015-06-03 07:58:23 <AmatCoder>     required by: mesa-dri-nouveau-10.5.6-r2[so:libLLVM-3.6.so]
2015-06-03 07:58:26 <ncopa> AmatCoder: which arch?
2015-06-03 07:58:32 <ncopa> edge?
2015-06-03 07:58:38 <AmatCoder> x86_64
2015-06-03 07:58:38 <ncopa> or v3.2
2015-06-03 07:58:49 <AmatCoder> 3.2 edge
2015-06-03 07:59:03 <ncopa> does 'apk version' give you any package names that needs upgrade?
2015-06-03 07:59:22 <AmatCoder> Installed:                                Available:
2015-06-03 07:59:27 <AmatCoder> busybox-1.23.2-r0                       < 1.23.2-r2
2015-06-03 07:59:36 <ncopa> only that?
2015-06-03 07:59:40 <AmatCoder> yes
2015-06-03 08:00:02 <ncopa> apk version -l '?'
2015-06-03 08:00:11 <ncopa> does that give anything?
2015-06-03 08:00:53 <AmatCoder> nop
2015-06-03 08:01:24 <ncopa> what do you have in /etc/apk/repositories?
2015-06-03 08:02:14 <AmatCoder> '/media/usb/apks'
2015-06-03 08:02:24 <AmatCoder> 'http://nl.alpinelinux.org/alpine/edge/main'
2015-06-03 08:02:34 <AmatCoder> 'http://nl.alpinelinux.org/alpine/edge/testing'
2015-06-03 08:02:45 <AmatCoder> not more
2015-06-03 08:04:10 <ncopa> and apk update does not fix it?
2015-06-03 08:04:11 <ncopa> hum
2015-06-03 08:04:16 <AmatCoder> 'llvm-libs-3.6.1-r1.apk' has not a 'libLLVM-3.6.so' file
2015-06-03 08:04:31 <ncopa> ncopa-desktop:~$ apk info -L llvm-libs
2015-06-03 08:04:31 <ncopa> llvm-libs-3.6.0-r1 contains:
2015-06-03 08:04:31 <ncopa> usr/lib/libLLVM-3.6.so
2015-06-03 08:04:31 <ncopa> usr/lib/libLTO.so
2015-06-03 08:04:31 <ncopa> usr/lib/BugpointPasses.so
2015-06-03 08:05:45 <AmatCoder> this is 3.6.0
2015-06-03 08:05:52 <AmatCoder> here it is 3.6.1
2015-06-03 08:06:44 <ncopa> aha
2015-06-03 08:06:45 <ncopa> ok
2015-06-03 08:07:51 <ncopa> argh
2015-06-03 08:08:02 <ncopa> so the 3.6.1 upgrade broke things
2015-06-03 08:08:09 <ncopa> and i even pushed it to v3.2
2015-06-03 08:08:24 <ncopa> i should never started with llvm 3.6 for alpine v3.2
2015-06-03 08:12:11 <ncopa> https://llvm.org/bugs/show_bug.cgi?id=15493
2015-06-03 08:12:25 <ncopa> with 3.6.1 we moved from ./configure to auto tools
2015-06-03 08:12:28 <ncopa> to cmake i mean
2015-06-03 08:13:55 <AmatCoder> I see
2015-06-03 08:16:34 <ncopa> i think a simple rebuild of mesa should fix it
2015-06-03 08:16:39 <ncopa> i'm on it
2015-06-03 08:17:16 <AmatCoder> thanks :)
2015-06-03 08:55:49 <AmatCoder> mesa issue is fixed...great!
2015-06-03 08:58:00 <ncopa> sorry about that
2015-06-03 08:58:24 <AmatCoder> not problem ;)
2015-06-03 09:16:34 <[xming]> I think this ARM board is very interesting http://wiki.t-firefly.com/index.php/Firefly-RK3288/en
2015-06-03 09:16:51 <[xming]> I might get one this summer when I am in Taiwan
2015-06-03 09:17:02 <[xming]> it has $GB RAM \o/
2015-06-03 09:17:06 <[xming]> 4GB*
2015-06-03 09:21:04 <[xming]> ncopa: is it possible to include custom js on the alpine wiki?
2015-06-03 09:23:11 <ncopa> [xming]: i doubt
2015-06-03 09:23:20 <ncopa> i as a wiki editor you mean?
2015-06-03 09:23:26 <ncopa> or you want add a wiki plugin?
2015-06-03 09:23:40 <[xming]> ncopa: nah, someone with fs access to put custom js
2015-06-03 09:24:02 <ncopa> i suppose it is possible
2015-06-03 09:24:25 <[xming]> if you are willing to put a few script, then I can make commmands clickable on the wiki and pops up a v86-alpine to execute the command
2015-06-03 09:24:39 <ncopa> you can create a ticket in infra tracker: http://bugs.alpinelinux.org/projects/alpine-infra/issues
2015-06-03 09:24:59 <ncopa> HA!
2015-06-03 09:25:13 <ncopa> sounds scary
2015-06-03 09:25:24 <ncopa> i mean
2015-06-03 09:25:50 <ncopa> what if i make a command that attacks someone else?
2015-06-03 09:25:54 <ncopa> then you click it and boom
2015-06-03 09:25:58 <[xming]> w/o network
2015-06-03 09:26:18 <[xming]> for the starter
2015-06-03 09:26:48 <ncopa> sounds cool though
2015-06-03 09:27:09 <[xming]> NIC is hardcoded in the js
2015-06-03 09:27:35 <[xming]> so w/o ws:// gateway there is "normally" no way a user can get out of his browser
2015-06-03 09:32:42 <[xming]> ncopa: I will put a demo I've done with "adv bash guide" o nthe tldp.org
2015-06-03 09:38:14 <kaniini> what is v86-alpine ?
2015-06-03 09:38:28 <coredumb> ncopa: If you've not noticed my taunt on #grsecurity
2015-06-03 09:38:56 <coredumb> would you be against providing a more hardened package of linux-grsec ?
2015-06-03 09:39:05 <coredumb> linux-hardgrsec maybe ?
2015-06-03 09:45:17 <kaniini> 
ACTION is

2015-06-03 09:45:26 <coredumb> :(
2015-06-03 09:45:37 <ncopa> kaniini: its an alpine in x86 emulator in javascript in browser
2015-06-03 09:45:47 <kaniini> if you want those features, you're better off building your own kernel
2015-06-03 09:45:56 <kaniini> because you can't trust that we didn't pre-own your kernel anyway
2015-06-03 09:46:14 <ncopa> coredumb: i think what we might want is a separate xen kernel
2015-06-03 09:46:42 <ncopa> so we can enable the interesting features we currently have disabled
2015-06-03 09:46:43 <kaniini> what i would prefer is a single APKBUILD that builds them all
2015-06-03 09:46:50 <ncopa> like kernexec
2015-06-03 09:46:50 <coredumb> ncopa: oh wait KERNEXEC and UDEREF are disabled because of Xen ?
2015-06-03 09:46:55 <kaniini> coredumb: yes
2015-06-03 09:46:58 <ncopa> coredumb: yes
2015-06-03 09:47:02 <ncopa> kaniini: we have that now...
2015-06-03 09:47:02 <coredumb> makes sense
2015-06-03 09:47:17 <kaniini> ncopa: i thought -vanilla was separate
2015-06-03 09:47:27 <ncopa> f70acf4e05206b0fddc040024f516b983a76418d
2015-06-03 09:47:48 <ncopa> we build multiple kernels from same patched kernel source
2015-06-03 09:48:03 <ncopa> so virtgrsec and grsec are build at same time
2015-06-03 09:48:06 <kaniini> but -vanilla is separate though
2015-06-03 09:48:07 <coredumb> and TPE :)
2015-06-03 09:48:11 <ncopa> -vanilla is different source
2015-06-03 09:48:13 <ncopa> different patch
2015-06-03 09:48:43 <ncopa> a xengrsec kernel would be built from same linux-grsec APKBUILD
2015-06-03 09:48:44 <kaniini> i guess what i am saying is, i would prefer that -grsec and -vanilla derive from a single root APKBUILD
2015-06-03 09:49:01 <ncopa> i am not sure we want that
2015-06-03 09:49:13 <ncopa> because config options might differ
2015-06-03 09:49:15 <kaniini> hmm, that is true
2015-06-03 09:49:18 <ncopa> also
2015-06-03 09:49:22 <ncopa> the price for joining them
2015-06-03 09:49:26 <kaniini> it's also bit debian in the ass too
2015-06-03 09:49:26 <rnalrd> vanilla and grsec also have different versions iirc
2015-06-03 09:49:45 <ncopa> if there comes an updated grsec patch
2015-06-03 09:49:51 <ncopa> for same base kernel version
2015-06-03 09:50:01 <ncopa> do we really want rebuild the vanilla, even if not needed?
2015-06-03 09:50:09 <kaniini> no what i am saying is that we have
2015-06-03 09:50:12 <kaniini> linux/APKBUILD
2015-06-03 09:50:15 <kaniini> linux-grsec/APKBUILD
2015-06-03 09:50:19 <kaniini> and linux-vanilla/APKBUILD
2015-06-03 09:50:28 <ncopa> linux/APKBUILD?
2015-06-03 09:50:35 <kaniini> linux/APKBUILD just pulls in the base source
2015-06-03 09:51:10 <kaniini> it doesnt actually build anything itself, it's just definitions for the others
2015-06-03 09:51:27 <ncopa> linus-sources
2015-06-03 09:51:28 <kaniini> then -grsec and -vanilla build the actual packages with their specialized patches + configs
2015-06-03 09:51:41 <kaniini> and then what you can do is
2015-06-03 09:51:46 <kaniini> apk add linux-sources
2015-06-03 09:51:53 <kaniini> and then build from that somehow
2015-06-03 09:51:56 <kaniini> with custom options
2015-06-03 09:51:59 <kaniini> using a custom APKBUILD
2015-06-03 09:52:37 <ncopa> so linux-vanilla would depend on linux-soruces
2015-06-03 09:52:48 <kaniini> yes
2015-06-03 09:53:00 <kaniini> and then we would have base kernelconfig's
2015-06-03 09:53:12 <kaniini> and then just patch them for the specific variants
2015-06-03 09:53:17 <ncopa> with linux-sources?
2015-06-03 09:53:19 <kaniini> yeah
2015-06-03 09:53:21 <kaniini> so that way
2015-06-03 09:53:31 <kaniini> linux-vanilla and linux-grsec both support the same drivers
2015-06-03 09:53:37 <ncopa> ok
2015-06-03 09:53:47 <ncopa> i see where you want go with that
2015-06-03 09:54:11 <kaniini> that way when you install -grsec or -hardgrsec or -customgrsec
2015-06-03 09:54:25 <kaniini> it's just a matter of specifying what you want different vs. stock
2015-06-03 09:54:47 <kaniini> this makes customizing the grsec knobs easier
2015-06-03 09:54:47 <ncopa> -grsec -rpi -whatever are just modifications of stock config
2015-06-03 09:54:59 <ncopa> we talked about that
2015-06-03 09:55:24 <ncopa> we want that even for the config-*.* in linux-grsec
2015-06-03 09:55:35 <ncopa> fedora as a config-common or similar
2015-06-03 09:55:46 <kaniini> so basically the source is always the same
2015-06-03 09:55:54 <kaniini> it's just different patches + configs
2015-06-03 09:56:04 <ncopa> which is the shared config options for all archs and all flavors
2015-06-03 09:56:05 <kaniini> so a 3.18.14 bug should expose itself on grsec and vanilla consistently
2015-06-03 09:56:45 <ncopa> *nod*
2015-06-03 09:57:09 <coredumb> well that wouldn't be _that_ consistent when grsec patch includes fixes that won't show up in upstream ...
2015-06-03 09:57:24 <ncopa> i am not sure we want do that with a shared linux-sources though
2015-06-03 09:57:37 <ncopa> mostly due to the different patching
2015-06-03 09:58:18 <ncopa> but a shared config would be nice
2015-06-03 09:58:31 <ncopa> maybe a kernel-baseconfig/APKBUILD
2015-06-03 09:58:47 <ncopa> which only holds the base config
2015-06-03 09:59:14 <ncopa> the vanilla could use an unmidfied config
2015-06-03 09:59:31 <ncopa> maybe with arch specific variants
2015-06-03 09:59:44 <ncopa> ok
2015-06-03 09:59:49 <ncopa> need to think about that
2015-06-03 09:59:54 <ncopa> not bad ideas
2015-06-03 10:00:10 <ncopa> other thing i am thinking of is a feature for abuild
2015-06-03 10:00:12 <kaniini> well we should probably avoid doing it the way debian did it
2015-06-03 10:00:54 <ncopa> i want support for multiple builds from same sources, with different configs
2015-06-03 10:01:02 <ncopa> out of tree builds that is
2015-06-03 10:01:09 <ncopa> php needs it
2015-06-03 10:01:13 <ncopa> zabbix needs it
2015-06-03 10:01:18 <ncopa> linux-grsec needs it
2015-06-03 10:01:31 <ncopa> most lua-* needs it
2015-06-03 10:02:24 <ncopa> in zabbix case we could have buildflavors="postgresql mysql sqlite3"
2015-06-03 10:02:35 <ncopa> then build() is called for each buildflavor
2015-06-03 10:02:37 <kaniini> coredumb: well grsec also includes bugs that doesn't show up in upstream :P
2015-06-03 10:02:39 <ncopa> or similar
2015-06-03 10:02:46 <coredumb> kaniini: indeed :)
2015-06-03 10:03:30 <coredumb> most of the time it's more like digging up crap of upstream though
2015-06-03 10:04:56 <kaniini> i'm irritated about that busybox setcap issue
2015-06-03 10:04:58 <ncopa> php's would be: buildflavors="cgi fpm apache2 embed"
2015-06-03 10:05:29 <ncopa> i think setcap behaviour is a bit unexpected
2015-06-03 10:05:44 <ncopa> ok i gotta get some lunch
2015-06-03 10:39:50 <[xming]> kaniini: http://xming.github.io/alpine-v86/demo2/instant.html
2015-06-03 10:43:28 <kaniini> nice
2015-06-03 10:54:20 <ScrumpyJack> ncopa: Mate is broken in testing du to llvm upgrade from 3.6 to 3.6.1
2015-06-03 10:54:52 <ncopa> ScrumpyJack: which of the packages?
2015-06-03 10:55:16 <ScrumpyJack> mate
2015-06-03 10:55:32 <ScrumpyJack>  testing/mate
2015-06-03 10:55:56 <ncopa> hum
2015-06-03 10:56:06 <ScrumpyJack> hmm, hang on
2015-06-03 10:56:09 <ncopa> i cannot find that....
2015-06-03 10:58:33 <ScrumpyJack> sorry, i meant mesa
2015-06-03 10:59:11 <ncopa> should be fixed with http://dup.pw/ddc8db1258f56b
2015-06-03 10:59:30 <ncopa> or does it not run?
2015-06-03 11:00:32 <ScrumpyJack> has that rebuilt in testing?
2015-06-03 11:00:41 <ScrumpyJack> perhaps it hasn't hit my mirrir
2015-06-03 11:01:03 <ncopa> it should be rebuilt in edge yes
2015-06-03 11:01:32 <ScrumpyJack> i still only see r2
2015-06-03 11:01:46 <ncopa> apk version -U
2015-06-03 11:01:48 <ncopa> does not show it?
2015-06-03 11:05:33 <ScrumpyJack> is it that problem with the web server responding to apk?
2015-06-03 11:05:55 <ScrumpyJack> the files on my mirror say r3 but the apk search says r2
2015-06-03 11:06:11 <ncopa> apk update?
2015-06-03 11:06:30 <ncopa> sounds like your local apkindex is not in sync
2015-06-03 11:06:47 <ScrumpyJack> yes, the apk update doesn't pull the new index
2015-06-03 11:06:59 <ScrumpyJack> because it's too new
2015-06-03 11:07:04 <ncopa> and apk update --force?
2015-06-03 11:07:09 <ScrumpyJack> apk update -f works
2015-06-03 11:07:21 <ncopa> might be a timezone issue then
2015-06-03 11:07:23 <ncopa> hum
2015-06-03 11:08:56 <ScrumpyJack> can you think why the package i installed on a usb stick dont install at boot time?
2015-06-03 11:09:12 <ncopa> dunno
2015-06-03 11:09:25 <ScrumpyJack> I have to run apk upgrade to get them from the usb stick
2015-06-03 11:19:52 <ncopa> kaniini: is there a way to print the Version: info for a given .pc file?
2015-06-03 11:20:42 <ncopa> i was thinking of adding: provides = <modulename>=<moduleversion>
2015-06-03 11:20:58 <ncopa> i was thinking of adding: provides = pc:<modulename>=<moduleversion>
2015-06-03 11:21:27 <ncopa> so we can do: depends = pc:modulename>=1.0 and similar
2015-06-03 11:21:45 <ncopa> and apk will handle version depends too
2015-06-03 11:23:06 <ncopa> but currently, i cannot se any other way than: version=$(awk '$1 == "Version:" { print $2}')
2015-06-03 11:25:00 <ncopa> hum
2015-06-03 11:25:04 <ncopa> that wont work either...
2015-06-03 11:25:08 <ncopa> znc/pkg/znc-dev/usr/lib/pkgconfig/znc.pc:Version: ${version}
2015-06-03 11:26:14 <ncopa> ah
2015-06-03 11:26:19 <ncopa> --modversion
2015-06-03 11:26:20 <ncopa> does it
2015-06-03 11:27:58 <vkrishn> zeromq at v4.1.1, testing/fossil at 1.33
2015-06-03 11:29:19 <kaniini> [xming]: v86 seems like a great tool for improving our docs
2015-06-03 11:41:56 <[xming]> kaniini: I am (thinking in my head) trying to make docs/tutorials more interactive in the UNIX/Linux CLI world
2015-06-03 11:42:19 <[xming]> kids these days all wants "web apps" :D
2015-06-03 11:42:25 <ncopa> :)
2015-06-03 11:42:29 <ncopa> true
2015-06-03 11:42:49 <[xming]> installing qmeu/virtualbox VM just for trying out 2 commands is a bit too complicated for most
2015-06-03 11:43:06 <kaniini> well, alpine is plumbing for "web apps" :P
2015-06-03 11:43:20 <[xming]> I want new comers to learn the true *nix way and not ubuntu windows alike way
2015-06-03 11:43:24 <kaniini> all this new stuff like docker and reproducible image mastering
2015-06-03 11:43:30 <kaniini> alpine was doing in 2009 :p
2015-06-03 11:44:33 <[xming]> was using BSD jail, solaris zones long before that
2015-06-03 11:45:10 <[xming]> personally I don't see any attraction in docker for me
2015-06-03 11:45:26 <kaniini> well i meant constructing an entire environment based on a specification file
2015-06-03 11:46:48 <[xming]> I know, the devs see that as great oppertunity and "end-users" too
2015-06-03 11:47:19 <[xming]> but as sysadmin/power user, this is really against our instinct :D
2015-06-03 11:47:51 <kaniini> its not, again, it's a typical pattern on alpine forever
2015-06-03 11:47:56 <kaniini> docker is just hype
2015-06-03 11:48:24 <kaniini> this is also hilarious, http://0pointer.net/blog/revisiting-how-we-put-together-linux-systems.html
2015-06-03 11:49:00 <ncopa> a way to do fast factory-reset
2015-06-03 11:49:02 <ncopa> heh
2015-06-03 11:49:39 <kaniini> if you have alpine configured for factory reset, you literally just powercycle the machine without committing the current image
2015-06-03 11:49:41 <kaniini> ;)
2015-06-03 11:50:00 <[xming]> grrr, we don't want that, we have btrfs for roll-backs :D
2015-06-03 11:50:06 <[xming]> or lvm snapshots
2015-06-03 11:50:21 <kaniini> well, that too.
2015-06-03 11:50:43 <[xming]> docker certain has its uses, but not for me (ATM)
2015-06-03 11:50:45 <kaniini> it would be interesting to add LVM snapshot integration to lbu(1).
2015-06-03 11:51:05 <kaniini> so you can get the same benefit without running from ram
2015-06-03 11:51:26 <kaniini> (but really many things should just run from RAM)
2015-06-03 12:03:32 <ahills> kaniini: that article made me really sad, thanks for ruining the first sunny day I've had up here
2015-06-03 12:12:13 <ncopa> kaniini: yes, i have thought about lvm snapshots too
2015-06-03 12:12:28 <ncopa> would be nice to be able to lvm snapshot, and apk upgrade
2015-06-03 12:12:34 <ncopa> and if everything ok, commit
2015-06-03 12:12:38 <ncopa> otherwise rollback
2015-06-03 12:12:49 <ncopa> ok
2015-06-03 12:12:59 <ncopa> seems like abuild handling pkg-config files works
2015-06-03 12:13:04 <ncopa> even with versions
2015-06-03 12:14:13 <kaniini> nice
2015-06-03 12:14:24 <ncopa> very nice actually
2015-06-03 12:14:35 <ncopa> we get version handling properly
2015-06-03 12:16:16 <ScrumpyJack> really can't get my head around why my usb won't install packages in cache
2015-06-03 12:16:21 <ScrumpyJack> from cache even
2015-06-03 12:23:58 <clandmeter> does it have the hidden file?
2015-06-03 12:24:52 <Jean-Scotch> ScrumpyJack: did you check the symlink in /etc/apk/cache ?
2015-06-03 12:25:54 <Jean-Scotch> I just had to fix it manually on my latest install
2015-06-03 12:26:24 <[xming]> I am so glad that kbus didn't make it this round
2015-06-03 12:26:45 <[xming]> kernel needs a more general msg API and not kbus specific
2015-06-03 12:27:09 <clandmeter> 
ACTION slaps ncopa with dup.pw

2015-06-03 12:28:41 <ncopa> 
ACTION slaps dup.pw with clandmeter

2015-06-03 12:29:04 <ncopa> i have an idea how to fix it
2015-06-03 12:29:10 <Jean-Scotch> and when the link was right, I had to issue apk update; apk cache -v sync. At next boot it was fixed.
2015-06-03 12:31:15 <vkrishn> any roadmap for v3.3, assuming its going to have some radical changes ?
2015-06-03 12:31:33 <barthalion> can we get abuild release?
2015-06-03 12:31:42 <ncopa> vkrishn: we will have better dependency detection for -dev packages
2015-06-03 12:31:51 <ncopa> so -dev packages will be more "optimized"
2015-06-03 12:31:51 <barthalion> I think my "remove *.la by default" isn't out yet
2015-06-03 12:32:06 <ncopa> barthalion: i think its in v3.2
2015-06-03 12:32:13 <ncopa> we probably need rebuild world though
2015-06-03 12:32:14 <barthalion> hm
2015-06-03 12:32:21 <ncopa> i think its there
2015-06-03 12:32:23 <barthalion> maybe it is, indeed
2015-06-03 12:32:38 <ncopa> the plan is test abuild a bit
2015-06-03 12:32:44 <ncopa> with the pkg-config fixes
2015-06-03 12:32:45 <kaniini> [xming]: well, a problem is that these people who say a "more general msg API is needed", have not really mocked up what one would look like
2015-06-03 12:32:49 <ncopa> and -dev fixes
2015-06-03 12:32:54 <ncopa> and then push gcc-5.1
2015-06-03 12:33:00 <vkrishn> so gcc 5.1, bettter -dev pkgs, ...
2015-06-03 12:33:03 <ncopa> and finally rebuild a whole lot of things...
2015-06-03 12:33:03 <ncopa> yes
2015-06-03 12:33:28 <ncopa> we might be able to get rid of default_dev in APKBUILDs
2015-06-03 12:33:35 <vkrishn> unified kernel ABUILD
2015-06-03 12:33:54 <ncopa> i want python3
2015-06-03 12:34:11 <kaniini> vkrishn: also unified raid management tools is something i am working on for 3.3
2015-06-03 12:35:28 <vkrishn> and APKBUILD script syntax changes ?
2015-06-03 12:35:43 <vkrishn> making it declarative ?
2015-06-03 12:39:31 <vkrishn> 7. improved builders, better mqtt msging
2015-06-03 12:40:56 <vkrishn> 8. better aports handling for java, go.. desktops ?
2015-06-03 12:41:30 <barthalion> changing APKBUILD is too big I guess, but it's up to ncopa as usually
2015-06-03 12:41:42 <clandmeter> 
ACTION slaps vkrishn with golang.org

2015-06-03 12:41:45 <vkrishn> rationale(8) , so that core building never gets stalled
2015-06-03 12:43:02 <vkrishn> nothing is stopping core devs in putting efforts in java, go, desktops apps
2015-06-03 12:43:12 <vkrishn> but it needs to be done my separate devs
2015-06-03 12:43:28 <barthalion> u w0t m8
2015-06-03 12:43:39 <BitL0G1c> we need ca-certificates-java for builds with maven to work
2015-06-03 12:43:54 <vkrishn> options are separate aports| domain
2015-06-03 12:44:58 <vkrishn> so goa-ports/(main|testing) will have only aports as deps
2015-06-03 12:45:02 <ncopa> i am not sure what to do with APKBUILD format
2015-06-03 12:45:05 <vkrishn> I mean go-aports
2015-06-03 12:45:17 <ncopa> just replacing it will probably be a too big change
2015-06-03 12:45:24 <coredumb> what's wrong with it ?
2015-06-03 12:45:34 <barthalion> ncopa: 1) let's start using set -e
2015-06-03 12:45:37 <vkrishn> yes, I guess, so maybe start a RFC in ml
2015-06-03 12:45:46 <barthalion> ncopa: 2) this is all
2015-06-03 12:45:46 <vkrishn> with proposed format
2015-06-03 12:45:47 <ncopa> barthalion: that we can do
2015-06-03 12:46:08 <barthalion> what is actually so limiting now?
2015-06-03 12:46:23 <ncopa> so limiting?
2015-06-03 12:46:35 <ncopa> with the APKBUILD format?
2015-06-03 12:46:41 <barthalion> well, you want to change it because of something
2015-06-03 12:47:08 <ncopa> i want be able to read all the dependencies for all subpackages
2015-06-03 12:47:11 <ncopa> in global scope
2015-06-03 12:47:45 <barthalion> what global scope means?
2015-06-03 12:47:53 <barthalion> "pkgbase" as we call it in Arch?
2015-06-03 12:47:58 <ncopa> not defined in a function
2015-06-03 12:48:06 <ncopa> so
2015-06-03 12:48:18 <ncopa> subpackages="foo"
2015-06-03 12:48:31 <ncopa> foo() { depends="bar"; }
2015-06-03 12:48:49 <ncopa> the depend for the subpackage foo is defined within the split function
2015-06-03 12:48:54 <ncopa> that is bad
2015-06-03 12:49:09 <coredumb> really ?
2015-06-03 12:49:11 <ncopa> as I need the depend for build time order
2015-06-03 12:49:22 <coredumb> oh
2015-06-03 12:49:39 <ncopa> how do i figure out *all* buildtime dependencies without actually running the splitfunctions
2015-06-03 12:49:50 <ncopa> mostly for build order
2015-06-03 12:49:55 <barthalion> I see now
2015-06-03 12:50:09 <ncopa> so, id like: foo.depends="bar"
2015-06-03 12:50:15 <ncopa> in global scope
2015-06-03 12:50:25 <ncopa> but shell does not support that
2015-06-03 12:50:28 <coredumb> that's not much of a change is it ?
2015-06-03 12:50:33 <coredumb> yeah indeed
2015-06-03 12:50:52 <ncopa> since shell does not support it we might need something else
2015-06-03 12:50:55 <barthalion> foo_depends?
2015-06-03 12:51:01 <ncopa> yeah, there are hacks
2015-06-03 12:51:21 <ncopa> there are also a couple of other limitations
2015-06-03 12:51:23 <clandmeter> bash :)
2015-06-03 12:51:31 <ncopa> bash does not solve all issues
2015-06-03 12:51:36 <ncopa> example
2015-06-03 12:51:36 <barthalion> XML
2015-06-03 12:51:38 <barthalion> 
ACTION hides

2015-06-03 12:51:39 <ncopa> lol
2015-06-03 12:51:43 <clandmeter> but its compatible
2015-06-03 12:51:45 <kaniini> rpm specfile
2015-06-03 12:51:51 <ncopa> pkgver=1.2.3
2015-06-03 12:51:51 <kaniini> ;)
2015-06-03 12:52:12 <ncopa> source="http://site/1.2/$pkgname-$pkgver"
2015-06-03 12:52:37 <ncopa> note the 1.2, the first 2 digits in the pkgver
2015-06-03 12:52:47 <ncopa> i do ${pkgver%.*}
2015-06-03 12:52:55 <ncopa> which basically cuts last digit
2015-06-03 12:53:06 <ncopa> but that fails as soon pkgver=1.2.3.1
2015-06-03 12:53:12 <ncopa> 4 digits
2015-06-03 12:53:25 <ncopa> so i need a way to split pkgver
2015-06-03 12:53:41 <ncopa> and pick N first digits
2015-06-03 12:54:45 <ncopa> ...without any forking (for performance)
2015-06-03 12:56:32 <ncopa> actually, I am seriously thinking of just using rpm spec file format
2015-06-03 12:56:41 <clandmeter> forking will delay reading that much?
2015-06-03 12:56:42 <BitL0G1c> ${pkgver#*.} - this catches all the decimal places
2015-06-03 12:56:43 <kaniini> it is pretty flexible
2015-06-03 12:57:06 <kaniini> ncopa: it'd also allow us to have more in-depth description fields
2015-06-03 12:57:16 <ncopa> yes forking slows down big times
2015-06-03 12:57:20 <kaniini> since its an rfc822-based format
2015-06-03 12:57:33 <kaniini> Description: foo dev
2015-06-03 12:57:37 <kaniini>   blah blah blah about foo
2015-06-03 12:57:41 <bmurphy1976> question.. I want to submit a new package (very small utility, zero dependencies, zero iteraction other tools, low probability of causing problems, etc.).  How does a package make the transition from testing to main?  Is there some sort of process or documentation somewhere explaining when a package goes where?
2015-06-03 12:57:42 <kaniini> is possible in specfile
2015-06-03 12:58:07 <kaniini> bmurphy1976: short story: packages in testing are transitioned when they have a maintainer and the maintainer has verified it works
2015-06-03 12:58:15 <ncopa> bmurphy1976: send patch to alpine-aports@lists.alpinelinux.org
2015-06-03 12:58:43 <ncopa> once its applied to testing, someone needs to test it
2015-06-03 12:58:50 <bmurphy1976> ok, so basically if I decide to be the maintainer it's up to me to guarantee that if it's in main it works as advertised?
2015-06-03 12:58:53 <ncopa> and verify that the build server's built version is ok
2015-06-03 12:59:00 <kaniini> bmurphy1976: maintainers are usually committers
2015-06-03 12:59:21 <ncopa> then we move it to main
2015-06-03 12:59:32 <bmurphy1976> ok, so basically I should submit to test and then somebody else would make the migration to main?
2015-06-03 12:59:37 <kaniini> yes
2015-06-03 12:59:58 <ncopa> yes, but nobody will move it to main unless someone can confirm the package is ok
2015-06-03 13:00:11 <bmurphy1976> ok, how do I go about making that confirmation?
2015-06-03 13:00:21 <ncopa> eg that it does not segfault
2015-06-03 13:00:24 <ncopa> on startup
2015-06-03 13:00:42 <kaniini> bmurphy1976: once the package is built write to alpine-devel requesting it be moved to main
2015-06-03 13:00:57 <ncopa> you add http://.../testing to your apk/etc/repositories
2015-06-03 13:01:03 <clandmeter> or ask here. most devs are here.
2015-06-03 13:01:06 <bmurphy1976> kaniini: ncopa: awesome, thank you, this is very helpful
2015-06-03 13:01:14 <ncopa> install it (the automatically built version, not then one you build yourself)
2015-06-03 13:01:15 <ncopa> test it
2015-06-03 13:01:21 <ncopa> and email, or ping us here
2015-06-03 13:01:28 <bmurphy1976> perfect, sounds reasonable enough
2015-06-03 13:01:29 <ncopa> or create an issue on bugs.a.o
2015-06-03 13:01:46 <kaniini> you can also just drop a patch here if you want to go that route
2015-06-03 13:02:25 <kaniini> the aports list is a good way to get your package committed for most people
2015-06-03 13:02:32 <kaniini> whatever works best for you
2015-06-03 13:02:41 <ncopa> yes. you can also do: git format-patch --stdout -1 | sprunge
2015-06-03 13:02:44 <ncopa> and paste url here
2015-06-03 13:02:52 <ncopa> specially if its a trivial package
2015-06-03 13:03:01 <ncopa> and if people here are not busy
2015-06-03 13:03:11 <ncopa> (we often are busy...)
2015-06-03 13:03:15 <clandmeter> :)
2015-06-03 13:03:25 <clandmeter> or we pretent
2015-06-03 13:03:35 <ncopa> i dont... :)
2015-06-03 13:03:47 <bmurphy1976> heh, ok, I think submitting to the mailing list is sufficient.  No need for me to get fancy here :)
2015-06-03 13:03:59 <ncopa> but i often pretend not being busy, and do blabla on irc.. :)
2015-06-03 13:04:19 <kaniini> bmurphy1976: what is the package anyway
2015-06-03 13:04:28 <ncopa> yeah im cureious too :)
2015-06-03 13:04:39 <bmurphy1976> parallel gzip: http://zlib.net/pigz/
2015-06-03 13:05:04 <ncopa> i thought gnu gzip did that too
2015-06-03 13:05:06 <bmurphy1976> we're switching some bootstrap tools from ubuntu to alpine because of the size and we make heavy use of it to speed up compression times
2015-06-03 13:06:09 <bmurphy1976> not sure, it's been some time since I looked at the state of the art and we built our tools, but last time around pigz was the best option for speeding up gzip
2015-06-03 13:06:26 <bmurphy1976> figured if I'm going to go through the trouble of building it, others might find it useful
2015-06-03 13:06:32 <kaniini> hmm, now i am curious
2015-06-03 13:06:58 <ncopa> send us the patch
2015-06-03 13:07:07 <bmurphy1976> will do!
2015-06-03 13:09:27 <vkrishn> what would be likely kernel version for v3.3 ?
2015-06-03 13:10:16 <vkrishn> 4.0.4 ?
2015-06-03 13:11:22 <fabled> cannot say yet
2015-06-03 13:11:37 <fabled> 3.3 is about 6 months away; and we'll look if grsecurity picks a new grsec-stable
2015-06-03 13:11:47 <fabled> but yes, probably something >=4.0
2015-06-03 13:11:53 <vkrishn> then...
2015-06-03 13:11:56 <kaniini> pigz seems to work
2015-06-03 13:12:01 <kaniini> i was bored and made an apkbuild :P
2015-06-03 13:12:04 <vkrishn> 9. kernel modules sfs splits
2015-06-03 13:13:12 <ncopa> :)
2015-06-03 13:13:13 <kaniini> bmurphy1976: i did it for you :p
2015-06-03 13:13:25 <bmurphy1976> lol
2015-06-03 13:13:33 <bmurphy1976> alright then, I was just about to submit :)
2015-06-03 13:13:39 <bmurphy1976> works for me
2015-06-03 13:13:50 <ncopa> looks nice
2015-06-03 13:13:56 <ncopa> pigz that is
2015-06-03 13:13:57 <ncopa> thanks
2015-06-03 13:14:07 <bmurphy1976> I'll give it a shot when it makes it out to the mirrors
2015-06-03 13:14:09 <bmurphy1976> thanks!
2015-06-03 13:16:34 <ncopa> vkrishn: tier1 mirrors syncing on mqtt messages is also somehint i'd like before v3.3
2015-06-03 13:16:43 <ncopa> and move git.a.o to gitolite
2015-06-03 13:17:10 <vkrishn> that would be '10'
2015-06-03 13:17:40 <ncopa> alpine-iso's makefile that generates releases
2015-06-03 13:17:49 <ncopa> needs some work too
2015-06-03 13:17:53 <ncopa> its way too fragile
2015-06-03 13:18:16 <bmurphy1976> thanks everybody, I need to head into the office but I'll definitely check later to make sure the package is working correctly
2015-06-03 13:18:18 <ncopa> might be i move the release scripts into aports git repo
2015-06-03 13:18:54 <ncopa> fabled: i'm buildling your gcc 5.1 patch here now
2015-06-03 13:19:00 <ncopa> on x86
2015-06-03 13:19:01 <fabled> ok
2015-06-03 13:19:05 <ncopa> i assume you tested x86_64
2015-06-03 13:19:08 <fabled> yes
2015-06-03 13:19:17 <fabled> just remembered, you might need to add isl to depends
2015-06-03 13:19:26 <ncopa> hum
2015-06-03 13:19:26 <ncopa> ok
2015-06-03 13:20:00 <ncopa> we shoudl add it to depends and a comment why it is there
2015-06-03 13:20:03 <BitL0G1c> ncopa - the rewrite I did of setup-disk with luks support also has uefi support - I'll put it on github so the changes are easier to see. The other script for custom installs 'setup-lvm' could be renamed 'setup-partitions' - this also has uefi support
2015-06-03 13:20:24 <ncopa> ah
2015-06-03 13:20:25 <ncopa> true
2015-06-03 13:20:30 <ncopa> ahills: ^^^
2015-06-03 13:20:36 <ncopa> have a look at BitL0G1c's stuff
2015-06-03 13:21:05 <ncopa> sha512sum: can't open '201-libitm.patch': No such file or directory
2015-06-03 13:21:05 <ncopa> 201-libitm.patch: FAILED
2015-06-03 13:21:08 <ncopa> fabled: ^^^
2015-06-03 13:21:47 <ncopa> http://sprunge.us/SNYS
2015-06-03 13:21:55 <fabled> huh
2015-06-03 13:22:21 <fabled> oh
2015-06-03 13:22:27 <ncopa> seems like the patch you sent me didnt include some of the patches
2015-06-03 13:22:36 <fabled> yes
2015-06-03 13:22:56 <fabled> try http://sprunge.us/PeFi
2015-06-03 13:23:02 <fabled> still missing isl
2015-06-03 13:24:04 <fabled> hmm
2015-06-03 13:24:30 <fabled> +  h = dlopen ("libisl.so.13", RTLD_LAZY);
2015-06-03 13:24:44 <fabled> so should probably depend on that
2015-06-03 13:32:44 <ncopa> depend="so:libisl.so.13" ?
2015-06-03 13:36:53 <vkrishn> any chance that hammer2, tux fs gets to v3.3
2015-06-03 13:36:54 <ahills> algitbot: are you crazy? sl takes like nothing to build!
2015-06-03 13:37:31 <ncopa> ahills: it needs ncurses-dev
2015-06-03 13:37:37 <ncopa> http://bld2.alpinelinux.org/buildlogs/build-edge-x86_64/testing/sl/sl-5.02-r0.log
2015-06-03 13:37:44 <vkrishn> ahills watch mqtt msgs. it shows firefox build in 1sec
2015-06-03 13:37:53 <ahills> ?
2015-06-03 13:38:15 <vkrishn> I mean if  you are familiar with msgs.a.o
2015-06-03 13:38:24 <ncopa> mosquitto_sub -h msg.alpinelinux.org -v -t 'build/#'
2015-06-03 13:41:55 <ncopa> fabled: that http://sprunge.us/PeFi is ok for push?
2015-06-03 13:42:07 <ncopa> or should it have an isl depend?
2015-06-03 13:42:31 <fabled> not sure, isl should probably added
2015-06-03 13:42:40 <fabled> i think
2015-06-03 13:42:44 <fabled> it should not depend on soname
2015-06-03 13:42:48 <fabled> since then you get the upgrade trouble
2015-06-03 13:42:52 <fabled> just depend on 'isl'
2015-06-03 13:42:54 <ncopa> depend=isl
2015-06-03 13:42:58 <fabled> and if the soname is right, it's used
2015-06-03 13:43:33 <ncopa> so i can upgrade isl first?
2015-06-03 13:43:37 <ncopa> or i need upgrade isl after
2015-06-03 13:44:37 <fabled> mmm
2015-06-03 13:44:47 <fabled> upgrading isl first, might current gcc stop work
2015-06-03 13:44:48 <ncopa> i think its safe to upgrade isl before
2015-06-03 13:45:05 <ncopa> there are no dep of isl in current gcc
2015-06-03 13:45:08 <fabled> oh
2015-06-03 13:45:09 <ncopa> so i think its totally unused
2015-06-03 13:45:09 <fabled> ok
2015-06-03 13:45:12 <fabled> then it's good
2015-06-03 13:45:34 <ncopa> http://sprunge.us/fMDg
2015-06-03 13:46:02 <ncopa> were thre more of gcc's deps that needed upgrade?
2015-06-03 13:46:50 <ncopa> or was it only isl
2015-06-03 13:50:18 <ncopa> i really like this abuild change
2015-06-03 13:50:35 <ncopa> its very nice to have the pkg-config handled
2015-06-03 13:50:57 <ahills> now it will be fun work to change every package's apkbuild to suit
2015-06-03 13:51:09 <ncopa> :)
2015-06-03 13:51:32 <ncopa> current apkbuilds will work
2015-06-03 13:51:52 <ahills> yes, but they won't be fashionable
2015-06-03 13:51:56 <ncopa> but it might make sense to rebuild from dependency order
2015-06-03 13:52:18 <ncopa> because if there is a provides=pc:module
2015-06-03 13:52:25 <ahills> ah yes
2015-06-03 13:52:30 <ncopa> then depend=pc:module
2015-06-03 13:52:49 <ncopa> so we want build packages from bottom i think
2015-06-03 13:52:59 <ahills> actually if there is an index of all files, searching for *.pc should not make it too difficult...
2015-06-03 13:53:11 <ncopa> the thing is
2015-06-03 13:53:15 <ncopa> i want push early
2015-06-03 13:53:33 <ncopa> then i want cleanup/rebuild along the way
2015-06-03 13:53:38 <ncopa> as we go
2015-06-03 13:53:39 <ncopa> however
2015-06-03 13:53:47 <ncopa> after new abuild is pushed
2015-06-03 13:54:06 <ncopa> if someone updates a package that has a dependency that is not rebuilt
2015-06-03 13:54:17 <ncopa> then this package will not be "prefect"
2015-06-03 13:54:25 <ncopa> hum
2015-06-03 13:54:36 <ncopa> and will need to be rebuilt
2015-06-03 13:54:43 <ncopa> i think that does not matter
2015-06-03 13:54:50 <ncopa> it will need be rebuilt anyways
2015-06-03 13:55:08 <ahills> well, the only time you have difficulty is when depends=pc:name and provides=pc:name is not in name/APKBUILD
2015-06-03 13:55:47 <ncopa> there are none of those
2015-06-03 13:55:58 <ncopa> provides=pc:name is automatically added
2015-06-03 13:56:11 <ahills> oh, so just the rebuild is needed then?
2015-06-03 13:56:15 <ncopa> yes
2015-06-03 13:56:18 <ncopa> well
2015-06-03 13:56:28 <ncopa> we have some depends_dev="name-dev"
2015-06-03 13:56:33 <ncopa> which will not be needed anymore
2015-06-03 13:56:57 <ncopa> we should problaby start move those to makedepends
2015-06-03 13:56:59 <ahills> will it be like depends vs. so detection, where it won't matter functionally?
2015-06-03 13:57:10 <ncopa> yes
2015-06-03 13:57:36 <ncopa> but
2015-06-03 13:57:39 <ncopa> with gcc-5.1
2015-06-03 13:57:59 <ncopa> we will probably want rebuild all packages using libstdc++ anyways
2015-06-03 13:58:31 <ncopa> so i think it makes sense to just rebuild world
2015-06-03 13:58:36 <ncopa> or as much as possible
2015-06-03 13:59:29 <ahills> BitL0G1c: could you do me a favor and ping me when you decide to publish setup-disk changes?
2015-06-03 13:59:33 <ahills> er, rewrite
2015-06-03 13:59:39 <ahills> I would like to try it out
2015-06-03 13:59:53 <ncopa> i saw some repo on github
2015-06-03 14:00:05 <ncopa> with some setup-disk changes with gpt support
2015-06-03 14:00:41 <ahills> I don't think I have any machines where I am forced to use uefi...
2015-06-03 14:03:09 <ncopa> hum
2015-06-03 14:03:15 <ncopa> when i think of depends_dev
2015-06-03 14:03:24 <ncopa> i think we want wait with cleaning up that
2015-06-03 14:03:39 <ncopa> at least to post v3.3
2015-06-03 14:04:08 <ncopa> we can let the depends_dev stay in APKBUILDs for now
2015-06-03 14:04:14 <ahills> when will new abuild become standard?
2015-06-03 14:04:21 <ncopa> mostly because it makes it easier to cherry-pick to v3.2
2015-06-03 14:04:38 <ncopa> ahills: i have new abuild in my dev env here
2015-06-03 14:04:45 <ncopa> i will build gcc 5.1
2015-06-03 14:04:50 <ahills> ok
2015-06-03 14:04:51 <ncopa> and build a bunch of packages
2015-06-03 14:04:55 <ncopa> to see what breaks
2015-06-03 14:04:59 <ahills> is gcc 5.1 some huge architecture change?
2015-06-03 14:04:59 <ncopa> if nothing breaks
2015-06-03 14:05:08 <ncopa> then i might push it soonish
2015-06-03 14:05:34 <ncopa> i was kind of hoping to rebuild world in my dev env...
2015-06-03 14:06:24 <ahills> I hope you have a powerful machine ;)
2015-06-03 14:06:42 <ncopa> i have 2 powerful machines ;)
2015-06-03 14:06:49 <ncopa> its the build servers
2015-06-03 14:06:57 <ahills> oh haha
2015-06-03 14:07:03 <ncopa> 48GB ram
2015-06-03 14:07:12 <ncopa> 24 cores ...
2015-06-03 14:07:27 <ncopa> build world would be super fast if it was not for autotools....
2015-06-03 14:08:12 <ahills> haha
2015-06-03 14:10:35 <ahills> oh my god... the mqtt official spec's authoritative version is .doc format...
2015-06-03 14:10:40 <ahills> how can anyone take them seriously?
2015-06-03 14:14:47 <ncopa> ok
2015-06-03 14:14:48 <vkrishn> ahills who/what company released it ?
2015-06-03 14:14:59 <ncopa> first build error with gcc-5.1
2015-06-03 14:15:03 <ncopa> ncurses
2015-06-03 14:15:13 <ahills> http://mqtt.org/documentation
2015-06-03 14:15:46 <vkrishn> http://mqtt.org/2014/11/mqtt-v3-1-1-now-an-oasis-standard
2015-06-03 14:17:17 <ahills> oasis' motto is "advancing open standards for the information society"
2015-06-03 14:17:27 <ahills> hilarious that they would use the most closed document format available
2015-06-03 14:21:00 <ncopa> barthalion: do arch linux have a patch for ncurses to build with gcc-5.1?
2015-06-03 14:21:45 <fabled> could we upgrade nl.a.o + fix the http headers to allow caching?
2015-06-03 14:23:50 <ncopa> nl.a.o upgraded
2015-06-03 14:24:26 <fabled> ncopa, which al?
2015-06-03 14:24:31 <ncopa> v3.2
2015-06-03 14:24:37 <fabled> ok
2015-06-03 14:24:50 <ncopa> but keep in mind that there is a reverse nginx in front of it
2015-06-03 14:24:57 <fabled> oh
2015-06-03 14:25:12 <fabled> can we update that?
2015-06-03 14:25:23 <fabled> which al it is?
2015-06-03 14:25:57 <ncopa> echo -n "GET / HTTP 1.1\r\n\r\n" | nc localhost 80
2015-06-03 14:26:04 <ncopa> how do i get him to print headers?
2015-06-03 14:26:10 <fabled> curl -D -
2015-06-03 14:26:35 <ncopa> HTTP/1.1 200 OK
2015-06-03 14:26:35 <ncopa> Content-Type: text/html
2015-06-03 14:26:35 <ncopa> Content-Length: 1507
2015-06-03 14:26:35 <ncopa> Date: Wed, 03 Jun 2015 14:26:18 GMT
2015-06-03 14:26:35 <ncopa> Server: lighttpd/1.4.35
2015-06-03 14:27:04 <ncopa> does it need provide any allow cachin headers?
2015-06-03 14:27:58 <fabled> i think it should give Last-Modified
2015-06-03 14:28:18 <fabled> that should come from backend
2015-06-03 14:28:19 <BitL0G1c> ahills - will put the 2 scripts on github today & ping you later on
2015-06-03 14:28:25 <ncopa> http://redmine.lighttpd.net/issues/1236
2015-06-03 14:29:17 <ahills> BitL0G1c: thanks! I'll be in the mountains today but this evening hopefully will have a chance to dig in
2015-06-03 14:29:24 <fabled> ha
2015-06-03 14:29:25 <fabled> i see
2015-06-03 14:29:47 <ncopa> HTTP/1.1 200 OK
2015-06-03 14:29:47 <ncopa> Content-Type: application/x-gzip
2015-06-03 14:29:47 <ncopa> Accept-Ranges: bytes
2015-06-03 14:29:47 <ncopa> ETag: "2043620334"
2015-06-03 14:29:47 <ncopa> Last-Modified: Wed, 03 Jun 2015 13:37:10 GMT
2015-06-03 14:29:48 <ncopa> Content-Length: 687778
2015-06-03 14:29:50 <ncopa> Date: Wed, 03 Jun 2015 14:29:26 GMT
2015-06-03 14:29:51 <ncopa> Server: lighttpd/1.4.35
2015-06-03 14:29:55 <fabled> smells that
2015-06-03 14:30:13 <fabled> maybe add that default mime-type then?
2015-06-03 14:30:21 <ncopa> yeah
2015-06-03 14:30:29 <ncopa> but with APKINDEX.tar.gz
2015-06-03 14:30:32 <ncopa> it looks correct
2015-06-03 14:30:34 <fabled> yes
2015-06-03 14:30:38 <fabled> it knows .gz
2015-06-03 14:30:43 <fabled> but not .apk, or .pkgdiff
2015-06-03 14:30:46 <fabled> or .xdelta
2015-06-03 14:30:49 <ncopa> # curl --silent --head localhost/alpine/edge/main/x86
2015-06-03 14:30:49 <ncopa> _64/APKINDEX.tar.gz | sprunge
2015-06-03 14:30:49 <ncopa> http://sprunge.us/gPaC
2015-06-03 14:31:06 <ncopa> we add those to mime-types?
2015-06-03 14:31:25 <fabled> it's serving static data, i'd just add the default "*" -> application/octet-stream
2015-06-03 14:31:35 <fabled> or ""
2015-06-03 14:32:06 <ncopa> would probably be nice to have a proper mime-type for .apk?
2015-06-03 14:32:07 <fabled> but if you wanna be fancy, do .apk, .iso and .xdelta at least
2015-06-03 14:32:12 <ncopa> yeah
2015-06-03 14:32:44 <ncopa> ".apk" => "application/x-tgz",
2015-06-03 14:32:45 <ncopa> ?
2015-06-03 14:36:19 <ncopa> # curl --silent --head localhost/alpine/edge/main/x86
2015-06-03 14:36:19 <ncopa> _64/zsh-5.0.7-r0.apk | sprunge
2015-06-03 14:36:19 <ncopa> http://sprunge.us/FRSS
2015-06-03 14:36:22 <ncopa> ok?
2015-06-03 14:37:18 <ncopa> clandmeter: i upgraded nl.a.o (alpine-mirror) to v3.2 and added mime-types for .apk, .iso and .xdelta
2015-06-03 14:37:22 <ncopa> just fyi
2015-06-03 14:40:46 <BitL0G1c> libreoffice / mate icons - could it be something as simple as running 'gtk-update-icon-cache' in the post-install script ?
2015-06-03 14:41:02 <ncopa> ncurses goes boom in generated c code
2015-06-03 14:41:10 <ncopa> how i love generated C ...
2015-06-03 14:41:30 <ncopa> https://groups.google.com/forum/#!topic/osv-dev/XnJCZuUoqt8
2015-06-03 14:41:40 <ncopa> patch is some sed awk magic...
2015-06-03 14:41:53 <fabled> ncopa, sure
2015-06-03 14:42:09 <ncopa> fabled: done.
2015-06-03 14:42:22 <ncopa> gentoo to the rescue...
2015-06-03 14:42:25 <ncopa> https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/sys-libs/ncurses/files/ncurses-5.9-gcc-5.patch
2015-06-03 14:46:26 <ncopa> and now a fortify issue
2015-06-03 14:46:38 <ncopa> In file included from ../../ncurses/curses.priv.h:84:0,
2015-06-03 14:46:38 <ncopa>                  from ../../ncurses/tty/hardscroll.c:148:
2015-06-03 14:46:38 <ncopa>   return __orig_snprintf(s, n, fmt, __builtin_va_arg_pack());
2015-06-03 14:46:39 <ncopa>          ^
2015-06-03 14:47:07 <clandmeter> ncopa: ok
2015-06-03 14:48:39 <fabled> ncopa, you are re-building world now?
2015-06-03 14:48:49 <ncopa> yes
2015-06-03 14:48:58 <ncopa> i started with build-base deps
2015-06-03 14:49:10 <ncopa> ncurses is not happy
2015-06-03 14:49:34 <ncopa> looks like ncurses is doing weird stuff with va_copy
2015-06-03 14:56:10 <ncopa> stateless: any idea why this happens with gcc-5 and latest-fortify? http://sprunge.us/VcVO
2015-06-03 14:56:13 <ncopa> it is ncurses
2015-06-03 14:59:42 <stateless> hmm
2015-06-03 15:00:23 <ncopa> i wonder if gcc-5 or somethign defines or  uses begin_va_copy?
2015-06-03 15:03:52 <ncopa> it fails on: #define begin_va_copy(dst,src)  va_copy(dst, src)
2015-06-03 15:03:52 <ncopa> #define end_va_copy(dst)        va_end(dst)
2015-06-03 15:08:19 <stateless> ncopa: what's CFLAGS?
2015-06-03 15:08:29 <stateless> can you try this one: http://sprunge.us/NFeT
2015-06-03 15:08:35 <stateless> requires at least -O1
2015-06-03 15:08:42 <ncopa> ccache gcc -DHAVE_CONFIG_H -I../ncurses -I../../ncurses -Os -fomit-frame-pointer -DNDEBUG -I. -I../include -I../../ncurses/../include -Os -fomit-frame-pointer  --param max-inline-insns-single=1200 -g -DTRACE -c ../../ncurses/tty/hardscroll.c -o ../obj_g/hardscroll.o
2015-06-03 15:08:46 <stateless> otherwise fails with the same error message (could be unrelated though)
2015-06-03 15:08:58 <stateless> should work with -Os
2015-06-03 15:09:14 <stateless> don't have a 5.1 gcc to test with here :(
2015-06-03 15:09:19 <vkrishn> clandmeter, what happens if someone ports linux kernel to go,
2015-06-03 15:09:20 <vkrishn> "golux", and if everybody stats booting it... hmm...
2015-06-03 15:09:20 <vkrishn> http://en.wikipedia.org/wiki/Revolution_%28TV_series%29
2015-06-03 15:09:46 <ncopa> $ gcc -O2 a.c
2015-06-03 15:09:47 <ncopa> ncdev-edge-musl-x86:~/test$
2015-06-03 15:09:52 <ncopa> worked
2015-06-03 15:10:09 <ncopa> ok
2015-06-03 15:10:18 <ncopa> failure does not happen with gcc-4.9.2
2015-06-03 15:10:43 <stateless> let me rework the test code
2015-06-03 15:10:50 <ncopa> i have to run
2015-06-03 15:11:39 <vkrishn> and we can release the super al-golux
2015-06-03 15:11:40 <ncopa> the gcc patch if you want test gcc-5.1: http://sprunge.us/LRTT
2015-06-03 15:11:52 <ncopa> ok see you tm
2015-06-03 15:12:41 <stateless> ncopa: what about this one? http://sprunge.us/GJUg
2015-06-03 15:13:10 <stateless> try with -Os
2015-06-03 15:49:02 <barthalion> ncopa: no
2015-06-03 16:24:13 <jomat> Hm... is it possible to add something with the effect of "set -e" to an APKBUILD? Adding || exit 1 after almost every line to successfully detect failed steps seems a bit cumbersome and error prone
2015-06-03 17:00:31 <ttilley> does busybox ash even support set -e ?
2015-06-03 19:04:10 <jomat> ttilley: Yes
2015-06-03 19:30:42 <clandmeter> can we have a abuild --check-deps-in-testing-when-i-want-to-move-it-to-main?
2015-06-03 19:47:33 <ahills> ttilley: every sh supports -e
2015-06-03 19:47:38 <ahills> that I have ever used
2015-06-03 19:47:44 <ahills> even ksh on old, old solaris
2015-06-03 19:48:08 <ahills> but to be more direct, yes, busybox ash supports it...
2015-06-03 20:07:02 <bmurphy1976> do I have to be subscribed to the aports mailing list to send a patch to it?
2015-06-03 20:08:51 <ahills> nope
2015-06-03 20:08:56 <ahills> but you might want to be, to see any replies ;)
2015-06-03 20:09:30 <ahills> not everyone's mail software works the same way
2015-06-03 20:09:59 <clandmeter> or check patchwork.a.o
2015-06-03 20:10:07 <bmurphy1976> yeah, I figured, but I sent a patch earlier today before I subscribed and I don't see it in the threaded web UI.  just trying to figure out if it made it or not.
2015-06-03 20:11:32 <bmurphy1976> I'm guessing the web ui refreshes infrequently
2015-06-03 20:12:22 <clandmeter> you mean patchwork?
2015-06-03 20:12:36 <ahills> bmurphy1976: it refreshes pretty frequently, just not every minute
2015-06-03 20:12:37 <clandmeter> or our lists.a.o?
2015-06-03 20:12:51 <bmurphy1976> lists.a.o
2015-06-03 20:12:59 <bmurphy1976> I set the patch a couple hours ago
2015-06-03 20:13:03 <clandmeter> it its a patch, it should be on patchwork
2015-06-03 20:13:26 <clandmeter> i pipe them directly with postfix
2015-06-03 20:13:53 <bmurphy1976> how do I submit it to patchwork?
2015-06-03 20:13:57 <clandmeter> i didnt setup lists.a.o, so i wouldnt know
2015-06-03 20:14:11 <clandmeter> patchwork == alpine-aports@
2015-06-03 20:14:17 <ahills> patchwork just reads alpine-aports
2015-06-03 20:14:32 <clandmeter> its subscribed to the ml
2015-06-03 20:14:33 <ahills> it will mail you when your patch is accepted/superseded/etc
2015-06-03 20:15:08 <ahills> bmurphy1976: I don't see your mail on the list, are you sure you sent it to the correct address?
2015-06-03 20:15:14 <bmurphy1976> I sent it to: alpine-aports@lists.alpinelinux.org
2015-06-03 20:15:19 <ahills> (assuming your name has "murphy" in it)
2015-06-03 20:15:28 <bmurphy1976> yes it does :D
2015-06-03 20:16:17 <ahills> sorry, I don't think it made it through, unless there is something clogging the list pipes
2015-06-03 20:16:51 <bmurphy1976> ok, I'll wait the usual 5 hours to see what the mail servers say, then try again this evening
2015-06-03 20:17:16 <barthalion> I don't see it either, so I guess it won't hit our servers
2015-06-03 20:17:38 <ahills> hopefully you're not graylisted
2015-06-03 20:17:57 <ahills> (sending from a blacklisted server, etc)
2015-06-03 20:18:14 <clandmeter> bmurphy1976: you can also sprunge it here
2015-06-03 20:18:19 <clandmeter> i can take a look at it
2015-06-03 20:21:18 <bmurphy1976> clandmeter: http://sprunge.us/QJch here you go
2015-06-03 20:30:02 <clandmeter> bmurphy1976: it doesnt need paxmarks anymore?
2015-06-03 20:30:47 <bmurphy1976> what exactly does paxmarks do?  I took the main/nginx package and copied the lua stuff into that.  That seemed to be the better maintained of the two packages.
2015-06-03 20:31:58 <clandmeter> its using luajit
2015-06-03 20:32:14 <clandmeter> grsec doesnt like jit
2015-06-03 20:32:26 <clandmeter> so we need to mark it
2015-06-03 20:32:38 <bmurphy1976> huh, ok, I'll redo the patch then and make sure it's still there
2015-06-03 20:32:49 <bmurphy1976> is there a way I can test this to verify it works?
2015-06-03 20:33:27 <clandmeter> i think its better if you send this to the list/patchwork, so others can look at it.
2015-06-03 20:33:41 <clandmeter> if you run a grsec kernel,  you can try to run it and see what happends.
2015-06-03 20:34:04 <ahills> what does paxmark do that paxctl doesn't?
2015-06-03 20:34:04 <bmurphy1976> ok, that's fine with me. I'll see if I can update my kernel and I'll figure out what is going on with my email.
2015-06-03 20:34:39 <clandmeter> ahills: ask fabled, he added it. i think he mentioned it before, but i forgot.
2015-06-03 20:34:55 <clandmeter> i just follow anything he does :)
2015-06-03 20:34:58 <ahills> haha
2015-06-03 20:36:43 <bmurphy1976> now I have another very strange problem
2015-06-03 20:36:49 <bmurphy1976> I created an alpine linux docker container with a small set of utilities installed
2015-06-03 20:36:53 <bmurphy1976> when I try to run a script I mounted into this container using a volume mapping, the script is cropped!
2015-06-03 20:36:57 <bmurphy1976> if I tail the file, I see the correct last few bytes, but if I cat the file, the last few bytes are missing (which is what the python interpreter sees)
2015-06-03 20:37:01 <bmurphy1976> this is very strange and I'm not sure if it's a docker problem, alpine linux problem, boot2docker issue, or something else
2015-06-03 21:02:43 <ahills> bmurphy1976: I'm not sure I understand what's happening... can you sprunge some terminal output?
2015-06-03 21:09:25 <bmurphy1976> ahills: http://sprunge.us/jWHV
2015-06-03 21:09:43 <bmurphy1976> you can see the "age])\r\n" is missing when I run the second command
2015-06-03 21:13:59 <bmurphy1976> I deleted the file, created a new one, pasted the same contents and now it is fine
2015-06-04 00:34:34 <shreeve> would like to create some alpine packages for some rubygems -- what is the minimal set of alpine packages to compile / create new apk's?
2015-06-04 00:49:47 <ttilley> shreeve: i don't know about minimal, but you probably want to just install alpine-sdk
2015-06-04 00:50:37 <ttilley> and then git, git checkout the ports tree, abuild-keygen, etc. check the wiki ;)
2015-06-04 00:51:46 <ttilley> i find a lot of projects tend to have stale and useless info all over the place, but alpine's wiki is actually pretty damn good. for the most part.
2015-06-04 01:17:33 <shreeve> thanks!!
2015-06-04 01:24:12 <shreeve> ttilley: the wiki was perfect, very helpful. more importantly, it all worked!
2015-06-04 01:24:29 <ttilley> awesome! glad to hear it :)
2015-06-04 02:25:08 <ahills> what is this "over quota" nonsense for sprunge?
2015-06-04 06:28:32 <barthalion> it's hosted on google app engine
2015-06-04 07:21:18 <ncopa> clandmeter: i am upgrading alpine-proxy to v3.2
2015-06-04 07:23:00 <clandmeter> go for it
2015-06-04 07:23:26 <clandmeter> ncopa: some gtk apps show weird (small) menu's. any idea what it can be?
2015-06-04 07:23:39 <ncopa> no?
2015-06-04 07:23:43 <ncopa> which gtk apps?
2015-06-04 07:23:50 <clandmeter> let me show you
2015-06-04 07:23:51 <ncopa> what gtk theme?
2015-06-04 07:23:54 <ncopa> have not seen it
2015-06-04 07:24:04 <clandmeter> i think im not using a theme
2015-06-04 07:24:52 <clandmeter> ncopa: https://dl.dropboxusercontent.com/u/18871200/Screenshot_2015-06-04_00-59-44.png
2015-06-04 07:25:14 <clandmeter> also when i go into settings, the tabs are really close to each other.
2015-06-04 07:25:29 <ncopa> can you try configure a theme?
2015-06-04 07:25:33 <ncopa> set up adwaita or sometehing
2015-06-04 07:25:45 <ncopa> there might be some gtk compat package for adwaita
2015-06-04 07:26:01 <ncopa> adwaita-gtk2-theme-3.14.0-r0
2015-06-04 07:26:44 <clandmeter> ill give it a try when im home
2015-06-04 08:53:02 <ncopa> whatta
2015-06-04 10:18:36 <fabled> hmmh
2015-06-04 10:19:26 <fabled> must need libtool
2015-06-04 10:52:13 <ahills> ncopa: I only copied systmkor's APKBUILD for terminus-font and didn't do any other work but adding that one makedepend, sorry
2015-06-04 11:32:13 <ahills> ncopa: minor patch for abuild: http://sprunge.us/CXbP
2015-06-04 11:33:24 <ncopa> ahills: that probably makes sense yes :)
2015-06-04 11:33:49 <ncopa> can you please ping me in 30 mins about it?
2015-06-04 11:33:56 <ahills> sure thing!
2015-06-04 11:34:15 <ncopa> i have my abuild git tree dirty due to new bugs i made
2015-06-04 11:34:29 <ahills> the best kind of bugs
2015-06-04 11:34:49 <ahills> I can also send to alpine-devel if you want
2015-06-04 11:35:13 <ncopa> nah
2015-06-04 11:35:16 <ncopa> its trivial
2015-06-04 11:35:32 <ncopa> i just want commit the fix i just confirmed here
2015-06-04 11:58:08 <ahills> most of those dup.pw links take me to a "bad object id"
2015-06-04 11:58:21 <ahills> I guess cgit doesn't like the short hash?
2015-06-04 11:58:32 <ncopa> dup.pw is a stupid redirector
2015-06-04 11:58:39 <ahills> right, just to cgit
2015-06-04 11:58:45 <ncopa> which only points you to cgit/aports
2015-06-04 11:58:52 <ahills> ohh
2015-06-04 11:58:58 <ncopa> it was a hack
2015-06-04 11:59:08 <ahills> just add one more layer of hack
2015-06-04 11:59:11 <ahills> dup.pw/aports/*
2015-06-04 11:59:17 <ahills> dup.pw/abuild/* ...
2015-06-04 11:59:32 <ncopa> hum
2015-06-04 12:00:27 <ncopa> server {
2015-06-04 12:00:27 <ncopa>         listen  192.168.9.80:80;
2015-06-04 12:00:27 <ncopa>         server_name     dup.pw;
2015-06-04 12:00:27 <ncopa> #       location  ~^/([0-9a-f]+)$ {
2015-06-04 12:00:27 <ncopa>         location  ~^/(.*)$ {
2015-06-04 12:00:28 <ncopa>                 return 301 $scheme://git.alpinelinux.org/cgit/aports/commit/?id=
2015-06-04 12:00:30 <ncopa>         }
2015-06-04 12:00:32 <ncopa> }
2015-06-04 12:00:55 <ncopa> thats the nginx config
2015-06-04 12:01:19 <ncopa> yeah
2015-06-04 12:01:24 <ncopa> thanks for that tip
2015-06-04 12:01:28 <ncopa> brilliant :)
2015-06-04 12:02:29 <ncopa> http://dup.pw/abuild/da666e00
2015-06-04 12:03:21 <ahills> I'm so glad you were using nginx so it was easy
2015-06-04 12:03:53 <ahills> sorry to everyone for inspiring ncopa to break all earlier links
2015-06-04 12:04:19 <ahills> hopefully algitbot will learn the change too
2015-06-04 12:06:33 <ncopa> ahills: xcmenu fails to build:
2015-06-04 12:06:34 <ncopa> xcmenu.c:14:21: fatal error: xcb/xcb.h: No such file or directory
2015-06-04 12:06:34 <ncopa>  #include <xcb/xcb.h>
2015-06-04 12:06:34 <ncopa>                      ^
2015-06-04 12:06:50 <ncopa> i think you need xcb-something in makedepends
2015-06-04 12:07:05 <ahills> yep, let me rewrite that in my new build machine
2015-06-04 12:07:14 <ahills> instead of on my laptop full of X includes
2015-06-04 12:12:55 <ahills> looks like I also missed zlib.h
2015-06-04 12:13:30 <ncopa> yup :)
2015-06-04 12:13:42 <ncopa> >>> xcmenu*: Tracing dependencies...
2015-06-04 12:13:42 <ncopa>         so:libc.musl-x86_64.so.1
2015-06-04 12:13:42 <ncopa>         so:libxcb.so.1
2015-06-04 12:13:42 <ncopa>         so:libz.so.1
2015-06-04 12:13:42 <ncopa> >>> xcmenu*: Package size: 56.0 KB
2015-06-04 12:14:06 <ncopa> that gives a hint that we need libxcb-dev and zlib-dev
2015-06-04 12:14:44 <ncopa> so:libz.so.1 (almost?) always means that it needs zlib-dev
2015-06-04 12:15:22 <ncopa> so it is normally good to check that all so:* are listed in makedepends
2015-06-04 12:15:40 <ncopa> ok
2015-06-04 12:15:50 <ncopa> i have an example of APKBUILD format failure
2015-06-04 12:16:23 <ncopa> https://github.com/alpinelinux/aports/blob/master/main/gtk%2B2.0/APKBUILD
2015-06-04 12:16:35 <ncopa> https://github.com/alpinelinux/aports/blob/master/main/gtk%2B2.0/APKBUILD#L101
2015-06-04 12:16:52 <ahills> sorry for the delay, sent updated aport
2015-06-04 12:16:53 <ncopa> a subpackage depends on hicolor-icon-theme
2015-06-04 12:17:54 <ncopa> >>> librsvg: Analyzing dependencies...
2015-06-04 12:17:54 <ncopa> ERROR: unsatisfiable constraints:
2015-06-04 12:17:54 <ncopa>   hicolor-icon-theme (missing):
2015-06-04 12:17:54 <ncopa>     required by: gtk-update-icon-cache-2.24.25-r0[hicolor-icon-theme]
2015-06-04 12:17:54 <ncopa>   .makedepends-librsvg-0:
2015-06-04 12:17:55 <ncopa>     masked in: cache
2015-06-04 12:17:57 <ncopa>     satisfies: world[.makedepends-librsvg]
2015-06-04 12:17:59 <ncopa> i just got the above error
2015-06-04 12:18:03 <ahills> is it too naive to >> sed -n '/^\s\+depends=/p' << ?
2015-06-04 12:18:10 <ncopa> from a build world
2015-06-04 12:18:21 <ncopa> yeah
2015-06-04 12:18:44 <ncopa> thats not very robust
2015-06-04 12:19:00 <ahills> I think depends_[subpkgname] is probably the best without making the format extremely different
2015-06-04 12:19:34 <ncopa> to actually parse that we need to depend on eval
2015-06-04 12:19:37 <ncopa> and eval is evil
2015-06-04 12:19:43 <ncopa> so i'd rather not go there
2015-06-04 12:19:56 <ahills> it's not totally evil, it makes up for the shell not being a fully featured language
2015-06-04 12:20:42 <ahills> I guess it does open the door for unexpected behavior during abuild, with malicious/accidental variable contents
2015-06-04 12:21:09 <ahills> but only subpackages is evaluated really
2015-06-04 12:21:27 <ncopa> for i in $subpackages; do eval "all_deps=\"$all_deps \$depends_$i\""; done
2015-06-04 12:22:16 <ncopa> i would prefer avoid eval for that
2015-06-04 12:22:35 <ncopa> i suppose alternative option is something like:
2015-06-04 12:22:56 <ahills> depends="zlib subpkgname:gtk"
2015-06-04 12:23:01 <ahills> or reverse those
2015-06-04 12:23:06 <ncopa> subpkgdepends="$pkgname-foo:dep1 $pkgname-foo:dep2"
2015-06-04 12:23:33 <ahills> that could be very verbose with advanced subpackages
2015-06-04 12:24:14 <ncopa> fortunally we normally dont have many depends in subpackages
2015-06-04 12:24:26 <ahills> you could also do subpkgdepends="$pkgname-foo:dep1 dep2 dep3 $pkgname-bar:dep1 dep2 dep3"
2015-06-04 12:25:28 <ncopa> http://sprunge.us/VTLK
2015-06-04 12:26:36 <ncopa> git and zabbix are the most advanced i think
2015-06-04 12:27:00 <ahills> oops, didn't actually send the updated aport because I forgot to enter my mail password
2015-06-04 12:27:23 <ncopa> for xcmenu?
2015-06-04 12:27:25 <ahills> yes
2015-06-04 12:27:32 <ncopa> i already pushed it....
2015-06-04 12:27:36 <ahills> oh haha
2015-06-04 12:27:41 <ncopa> :)
2015-06-04 12:27:41 <ahills> did you add zlib-dev to makedepends?
2015-06-04 12:27:45 <ncopa> yes
2015-06-04 12:27:50 <ahills> thanks
2015-06-04 12:27:54 <ncopa> http://dup.pw/aports/6ccd7e7b
2015-06-04 12:28:34 <ncopa> those are the subpackages with more than a oneliner as depends: http://sprunge.us/OMeA
2015-06-04 12:29:11 <ahills> for complicated depends, there is nothing stopping the packager from using $()
2015-06-04 12:30:05 <ncopa> i would stop the packager from using $()
2015-06-04 12:30:08 <ahills> e.g. subpkgdepends="$(echo dep1 dep2 dep3 dep4 dep5 dep6 | sed 's/^/subpkg:/')"
2015-06-04 12:30:13 <ahills> why?
2015-06-04 12:30:20 <ncopa> performance
2015-06-04 12:30:22 <ahills> ah
2015-06-04 12:30:29 <ncopa> i parse all APKBUILDs
2015-06-04 12:30:32 <ncopa> from script
2015-06-04 12:30:38 <ahills> yep makes sense
2015-06-04 12:30:54 <ahills> dependencies are not so dynamic anyways
2015-06-04 12:31:10 <ncopa> zabbix has same problem as gtk+2.0...
2015-06-04 12:31:32 <ncopa> bootstrapping a repo might try build zabbix before php
2015-06-04 12:31:39 <ahills> are you going to add subpkgdepends?
2015-06-04 12:31:45 <ncopa> i dont know
2015-06-04 12:31:47 <ahills> I have some time today, can help fix up apkbuilds
2015-06-04 12:32:22 <ncopa> for now the workaround is to manually add that to makedepends as well
2015-06-04 12:32:25 <ahills> good thing apkbuild format already precludes subpkgmakedepends
2015-06-04 12:33:05 <ncopa> so in gtk+2.0 case, we can workaround it by adding hicolor-icon-theme to makedepends
2015-06-04 12:33:56 <ncopa> this is one of the issues i'd like to solve if we go for a completely new format
2015-06-04 12:38:47 <ncopa> i think we just workaround it for now
2015-06-04 12:53:21 <ahills> is there any c library that provides bsd's arc4random?
2015-06-04 12:55:10 <ncopa> ahills: i think uclibc provided it
2015-06-04 12:55:27 <ncopa> maybe libbsd does
2015-06-04 13:00:21 <ahills> looks like libbsd does, thanks
2015-06-04 13:10:47 <ncopa> $ ls *.apk | wc -l
2015-06-04 13:10:47 <ncopa> 727
2015-06-04 13:11:07 <ncopa> 727 .apk packages build with gcc-5.1 and new abuild
2015-06-04 13:14:34 <ahills> how many packages total? something like 14000?
2015-06-04 13:14:45 <ncopa> probably like 5000
2015-06-04 13:15:22 <ncopa> $ ls *.apk | wc -l
2015-06-04 13:15:22 <ncopa> 5270
2015-06-04 13:16:17 <ncopa> i think i'll push gcc5+abuild when it has built atleast 50%
2015-06-04 13:56:40 <ahills> are the two related or could they be pushed independently?
2015-06-04 13:58:49 <ncopa> they could ofcourse be pushed independently
2015-06-04 13:59:18 <ncopa> but we probably want push them close together as they both would benefit from a rebuild of world
2015-06-04 13:59:23 <ncopa> hum
2015-06-04 13:59:27 <ncopa> perl is broken
2015-06-04 13:59:56 <ncopa> $ echo "use IO::Socket;" | perl
2015-06-04 13:59:56 <ncopa> Bareword "Errno::EINVAL" not allowed while "strict subs" in use at /usr/lib/perl5/core_perl/IO/Socket.pm line 135.
2015-06-04 13:59:56 <ncopa> Compilation failed in require at - line 1.
2015-06-04 13:59:56 <ncopa> BEGIN failed--compilation aborted at - line 1.
2015-06-04 14:00:22 <fabled> gcc5.1 ?
2015-06-04 14:01:14 <fabled> or new perl?
2015-06-04 14:01:34 <ncopa> http://permalink.gmane.org/gmane.linux.lfs.devel/16115
2015-06-04 14:01:41 <ncopa> i think its gcc5 that causes it
2015-06-04 14:02:45 <fabled> ok
2015-06-04 14:02:48 <ncopa> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=778060
2015-06-04 14:03:00 <fabled> there was kernel issues too, i think they got mostly patched in -stables already
2015-06-04 14:03:54 <fabled> there's upstream fix. nice.
2015-06-04 14:53:12 <ncopa> t/op/magic .................................................... FAILED--unexpect
2015-06-04 14:53:12 <ncopa> ed output at test 102
2015-06-04 14:53:19 <ncopa> perl issue
2015-06-04 16:38:38 <clandmeter> who broke abuild?
2015-06-04 16:39:46 <clandmeter> hmm
2015-06-04 16:42:19 <clandmeter> i broke myself it seems
2015-06-04 16:43:02 <clandmeter> everytime i think i can blame ncopa, something comes up.
2015-06-04 18:06:15 <ahills> try /nick ncopa first, then make the mistake, then sneak back out of your costume
2015-06-04 18:18:12 <systmkor> ahills: I'll fix it up. Sorry about not finishing the terminus font.
2015-06-04 18:22:56 <ahills> fix which?
2015-06-04 18:23:11 <ahills> systmkor: no problem, I was just impatient, and your patch was already on the list
2015-06-04 18:23:36 <systmkor> the terminus-font. I'll fix the APKBUILD right now :D
2015-06-04 18:32:27 <ahills> oh, I think ncopa already did
2015-06-04 18:32:53 <ahills> http://git.alpinelinux.org/cgit/aports/tree/testing/terminus-font/APKBUILD
2015-06-04 18:32:58 <ahills> systmkor: ^
2015-06-04 18:33:13 <systmkor> oh okay cool
2015-06-04 18:34:42 <systmkor> is it standard to us _at_ instead of @?
2015-06-04 18:35:12 <systmkor> in the maintainer & contributor area of the APKBUILD?
2015-06-04 18:36:26 <ahills> no I think that's cgit automatically doing it
2015-06-04 18:36:30 <ahills> it shows up as @ in my git clone
2015-06-04 18:37:08 <systmkor> ahhh m'kay
2015-06-04 18:38:33 <systmkor> ahills: needing to do some backup of data and re-install of alpine on my laptop
2015-06-04 18:38:44 <systmkor> ahills: after that excited to use testing/beard
2015-06-04 18:38:46 <systmkor> :D
2015-06-04 18:40:01 <ahills> :D
2015-06-04 18:40:07 <ahills> feedback welcome!
2015-06-04 18:40:37 <ahills> I'm actually currently not using it because the grsec kernel does not allow hibernation, and I got distracted from my project to fix that
2015-06-04 18:40:52 <ahills> it's very simple, I'm getting hung up on packaging
2015-06-04 18:41:05 <systmkor> ahills: how does grsec prevent hibernation?
2015-06-04 18:41:49 <ahills> someone in here knows the details better than me, I think stf (currently aka bCQc0P3DIqbh)
2015-06-04 18:42:08 <ahills> but basically because dumb users would hibernate to unencrypted swapfiles, it's a config gate, and there's a simple patch to remove it
2015-06-04 18:42:24 <ahills> I'll actually get to that in fifteen minutes, I'm really interested in hibernating before I run out of battery today :P
2015-06-04 18:43:42 <systmkor> ahills: well I guess we should move to encrypted swap regardless
2015-06-04 18:44:01 <ahills> yes, I use a swapfile on an encrypted partition
2015-06-04 18:44:51 <ahills> but, it's very inconvenient to me, because I actually *do* use FDE, and also want to hibernate :P
2015-06-04 18:44:54 <bCQc0P3DIqbh> confirmed. and agreed, you should only swap to encrypted partitions.
2015-06-04 18:45:16 <systmkor> ahills: well it should be possible to do FDE and hibernate
2015-06-04 18:45:23 <systmkor> just would be more tricky
2015-06-04 18:45:25 <ahills> if you have cryptsetup installed on alpine, /etc/conf.d/dmcrypt has got proper instructions for managing encrypted swap automatically
2015-06-04 18:45:38 <ahills> systmkor: it's not difficult, I had it working on my arch install before I nuked that for alpine
2015-06-04 18:46:06 <systmkor> ah cool, yah that's what I want to do, going to try and do backup and re-install of Alpine on my laptop
2015-06-04 18:47:51 <ahills> which laptop you have?
2015-06-04 18:48:26 <systmkor> ThinkPad T510
2015-06-04 18:49:52 <systmkor> oh quick question when I have errors being reported by apk, how do I get apk to specify which packages or files are causing said errors?
2015-06-04 18:54:53 <ahills> what do you mean?
2015-06-04 18:55:06 <ahills> oh, if it reports at the end and you've lost your scrollback?
2015-06-04 18:55:14 <ahills> apk fix tries to fix them, and should be telling you what it's doing
2015-06-04 19:07:33 <systmkor> ah, i thought apk fix required a specific package
2015-06-04 19:07:34 <systmkor> thanks
2015-06-04 19:33:28 <ahills> What does it mean when -Wl,--as-needed causes a build to fail with undefined references?
2015-06-04 19:34:22 <ahills> oh, this is probably just an order of arguments problem
2015-06-04 20:08:54 <barthalion> yes, it's usually ordering or missing linker flags
2015-06-05 06:13:08 <ncopa> morning
2015-06-05 06:13:11 <ncopa> new musl. nice!
2015-06-05 06:13:21 <ncopa> fabled: i suppose this is safe to remove now:
2015-06-05 06:13:23 <ncopa> -       [ "$CLIBC" = "musl" ] && export CFLAGS="$CFLAGS -DNULL=0L"
2015-06-05 06:13:29 <ncopa> in openldap
2015-06-05 06:15:48 <fabled> if it works, yes
2015-06-05 06:17:13 <ncopa> ok, thanks
2015-06-05 06:20:07 <fabled> ncopa, new musl is mostly regression/bug fix only. so i'll cherry-pick that to 3.2-stable after a while
2015-06-05 06:20:13 <fabled> before 3.2.1
2015-06-05 06:20:41 <ncopa> ok
2015-06-05 06:21:11 <ncopa> file an issue on bugs.a.o with target 3.2.1 if you want make sure to not forget it
2015-06-05 06:21:47 <ncopa> seems like most issues with world rebuild are due to gcc-5
2015-06-05 06:46:46 <ncopa> new perl too
2015-06-05 06:47:07 <ncopa> i think i'll push gcc 5 today
2015-06-05 06:47:10 <ncopa> hopefully
2015-06-05 06:47:12 <ncopa> and new abuild
2015-06-05 06:58:19 <fabled> new musl + geany crashes
2015-06-05 06:58:24 <fabled> could be application error
2015-06-05 06:58:38 <fabled> strerror_l called with locale=null
2015-06-05 06:58:42 <fabled> i think that's not really legal
2015-06-05 06:59:23 <ncopa> ask musl ppl
2015-06-05 06:59:35 <ncopa> they are the standards lawyers ;)
2015-06-05 07:00:21 <fabled> i did
2015-06-05 07:09:36 <ncopa> main total built:       1758
2015-06-05 07:09:36 <ncopa> main total relevant aports:     2034
2015-06-05 07:09:55 <ncopa> the big majority builds with gcc-5 and new abuild
2015-06-05 07:15:35 <fabled> cool
2015-06-05 07:15:50 <fabled> i'm fixing one issue in strongswan
2015-06-05 07:15:56 <fabled> please let me push that first
2015-06-05 07:16:06 <fabled> i'm testing it now
2015-06-05 07:16:09 <ncopa> i will no push til afternoon i think
2015-06-05 07:16:10 <fabled> ok
2015-06-05 07:16:16 <ncopa> i need tag release of abuild first
2015-06-05 07:16:22 <ncopa> i am adding some warnings
2015-06-05 07:16:31 <ncopa> >>> WARNING: libtheora-dev*: Could not find any provider for pc:ogg>=1.1
2015-06-05 07:16:32 <ncopa> >>> WARNING: libtheora-dev*: libogg-dev needs to be rebuilt
2015-06-05 07:30:05 <fabled> argm
2015-06-05 07:30:13 <fabled> new musl breaks also xfce-terminal
2015-06-05 07:30:22 <ncopa> :-(
2015-06-05 07:30:24 <fabled> sounds like the issue is in some library
2015-06-05 07:31:34 <fabled> probably all caused by that strerror_l misuse
2015-06-05 07:31:41 <fabled> i'll add a hacky fixfor that
2015-06-05 07:31:43 <fabled> orsimilar
2015-06-05 07:32:05 <ncopa> and possibly spank dalias
2015-06-05 07:36:40 <fabled> hmm
2015-06-05 07:36:41 <fabled> no
2015-06-05 07:36:43 <fabled> it's something else
2015-06-05 07:40:20 <fabled> i think i found it
2015-06-05 07:46:51 <fabled> final test, and i'll push fix
2015-06-05 09:03:20 <ncopa> ok
2015-06-05 09:03:27 <ncopa> new issues with gcc5
2015-06-05 09:03:44 <ncopa> argp-standalone does not build
2015-06-05 09:05:04 <ncopa> oh
2015-06-05 09:05:23 <ncopa> we shoudl try make the eudev move too relatively soon after gcc5 i suppose
2015-06-05 09:05:31 <ncopa> as that also means a bigger rebuild
2015-06-05 09:26:39 <ncopa> (115/122) Installing gtk+2.0 (2.24.25-r0)
2015-06-05 09:26:39 <ncopa> Executing gtk+2.0-2.24.25-r0.post-install
2015-06-05 09:26:39 <ncopa> Segmentation fault (core dumped)
2015-06-05 09:26:42 <ncopa> hum
2015-06-05 09:29:01 <ncopa> ok that segfault was due to the musl thingy
2015-06-05 09:37:13 <ncopa> ok, i have one issue with abuild
2015-06-05 09:37:59 <ncopa> problem happens when you have one subpackage with a symlink usr/lib/libfoo.so pointing to: ../../lib/libfoo.so.0
2015-06-05 09:38:10 <ncopa> which is provided by another package (foo-libs)
2015-06-05 09:38:26 <ncopa> and that package does not have usr/lib
2015-06-05 09:39:06 <ncopa> so basically, usr/lib/../../lib/libfoo.so.0 is invalid when usr/lib does not exist, even if lib/libfoo.so.0 does
2015-06-05 09:39:44 <ncopa> and idea how to convert usr/lib/../../lib/libfoo.so.0 to lib/libfoo.so.0 ?
2015-06-05 09:50:57 <ncopa> stateless: nss fails with fortify-headers (git) and gcc5:
2015-06-05 09:51:31 <vkrishn> ncopa, I wanted to ask a similar query for libfoo.so.0 issue
2015-06-05 09:51:35 <ncopa> http://sprunge.us/aXDi
2015-06-05 09:52:05 <vkrishn> shouldn't symlinks to lib file pkged in <pkg>-dev be in the libpkg itself?
2015-06-05 09:52:25 <ncopa> no
2015-06-05 09:52:39 <ncopa> why should it?
2015-06-05 09:53:26 <vkrishn> I don't know for sure
2015-06-05 09:53:54 <vkrishn> eg, http://pkgs.alpinelinux.org/contents?pkgname=libzmq&arch=x86
2015-06-05 09:54:03 <vkrishn> usr/lib/libzmq.so.4.0.0
2015-06-05 09:54:09 <vkrishn> usr/lib/libzmq.so.4 -> libzmq.so.4.0.0
2015-06-05 09:54:18 <vkrishn> BUT, http://pkgs.alpinelinux.org/contents?pkgname=zeromq-dev&arch=x86
2015-06-05 09:54:22 <vkrishn> has
2015-06-05 09:54:29 <vkrishn> usr/lib/libzmq.so -> libzmq.so.4.0.0
2015-06-05 09:56:24 <ncopa> yes
2015-06-05 09:56:27 <ncopa> that is correct
2015-06-05 09:56:56 <ncopa> you only need libzmq.so build time
2015-06-05 09:56:59 <ncopa> during linking
2015-06-05 09:57:11 <ncopa> gcc -lzmq ... needs it
2015-06-05 09:57:36 <ncopa> if you do: gcc -o myprog -lzmq ...
2015-06-05 09:57:46 <ncopa> and readelf -d myprog
2015-06-05 09:57:58 <ncopa> you will have a NEEDED libzmq.so.4
2015-06-05 09:58:23 <ncopa> during runtime, the dynamic loader will only need libzmq.so.4
2015-06-05 09:58:47 <ncopa> you can also do readelf -d usr/lib/libzmq.so.4.0.0
2015-06-05 09:59:09 <ncopa> and it should show SONAME libzmq.so.4
2015-06-05 09:59:33 <ncopa> so libzmq.so should be in -dev package only
2015-06-05 09:59:46 <ncopa> why would you need it in the runtime?
2015-06-05 10:00:53 <vkrishn> thought if it points to same file , it should be in libpkg
2015-06-05 10:01:07 <vkrishn> but, thanks for explanation
2015-06-05 11:07:38 <[xming]> more v86 demo http://xmingske.bitbucket.org/abs.html
2015-06-05 11:08:57 <[xming]> command are double clickable
2015-06-05 11:09:09 <[xming]> and will be send to the terminal on the right
2015-06-05 11:09:47 <[xming]> not everything will work as we only have ash
2015-06-05 11:36:32 <ScrumpyJack> xorg-server depends on wayland. can we remove this dependancy or have I missed something
2015-06-05 11:37:10 <ScrumpyJack> also, i thought apk del -f would allow me to remove a package ignoring dependancies. is that correct?
2015-06-05 11:58:30 <fcolista> how is it possible to check the alpine version if i've only the kernel version with uname -ar ?
2015-06-05 11:58:41 <fcolista> do we have any association?
2015-06-05 12:04:07 <ScrumpyJack>  what's the kernel version you have?
2015-06-05 12:04:34 <clandmeter> fcolista: check git log?
2015-06-05 12:05:15 <fcolista> uname -ar returns 2.6.32.16-grsec #1-Alpine SMP Wed Jul 7 13:17:21 UTC 2010 i686 GNU/Linux
2015-06-05 12:05:25 <fcolista> it should be 1.10
2015-06-05 12:07:20 <ScrumpyJack> would do a find but that's too old for my mirror
2015-06-05 12:09:54 <clandmeter> http://git.alpinelinux.org/cgit/aports/commit/main?h=1.10-stable&id=161481a751ba1bbe58ac86c08d4507e2afb1eaa1
2015-06-05 12:10:03 <clandmeter> thats 2.6.32.16
2015-06-05 12:13:18 <fcolista> thx clandmeter
2015-06-05 12:14:00 <ahills> fcolista: check /etc/os-release I think? Not sure what version that was added
2015-06-05 12:14:23 <fcolista> /etc/alpine-version was not available
2015-06-05 12:14:52 <fcolista> nor os-release
2015-06-05 12:15:53 <clandmeter> probably added later on
2015-06-05 12:19:36 <clandmeter> fcolista: v2.0.0_beta3
2015-06-05 12:20:17 <fcolista> thx
2015-06-05 12:21:37 <clandmeter> or any beta/rc afterwards ;-)
2015-06-05 12:27:51 <ncopa> hum
2015-06-05 12:28:02 <ncopa> nspr does not have any .so symlinks
2015-06-05 12:28:28 <ncopa> so autodetection of nspr-dev's depend fails
2015-06-05 12:28:39 <ncopa> it fails to detect that it needs depend=nspr
2015-06-05 12:29:49 <ncopa> $ PKG_CONFIG_PATH=pkg/nspr-dev/usr/lib/pkgconfig pkg-config --libs nspr
2015-06-05 12:29:49 <ncopa> -lplds4 -lplc4 -lnspr4
2015-06-05 12:30:09 <ncopa> i wonder if i should make abuild try find -l*
2015-06-05 12:30:14 <ncopa> to autodetect deps
2015-06-05 12:34:51 <ncopa> >>> libxfce4ui: Analyzing dependencies...
2015-06-05 12:34:51 <ncopa> ERROR: unsatisfiable constraints:
2015-06-05 12:34:51 <ncopa>   Huh? Error reporter did not find the broken constraints.
2015-06-05 13:15:08 <ahills> ncopa: if there's no symlink does it actually depend on it?
2015-06-05 13:15:23 <ncopa> no
2015-06-05 13:15:26 <ncopa> that is the problem
2015-06-05 13:15:33 <ncopa> well
2015-06-05 13:15:34 <ncopa> yes
2015-06-05 13:15:36 <ncopa> it does
2015-06-05 13:15:53 <ncopa> if pkg-config --libs give you: -lfoo
2015-06-05 13:15:57 <ahills> ah yes
2015-06-05 13:16:05 <ncopa> and libfoo.so is not a symlink
2015-06-05 13:16:30 <ncopa> i think I'll just push it as is for now
2015-06-05 13:16:49 <ncopa> we can manually manage the depends_dev for those execptions
2015-06-05 13:17:40 <ncopa> i think i want push new abuild and gcc5 today
2015-06-05 13:17:53 <ahills> exciting
2015-06-05 13:17:57 <ahills> then you can leave for the weekend ;)
2015-06-05 13:18:00 <ncopa> but it is risky doing so on a friday afternoon
2015-06-05 13:18:01 <ncopa> :)
2015-06-05 13:18:04 <ncopa> exactly
2015-06-05 13:18:21 <ncopa> i suppose it is more wise to wait til monday
2015-06-05 13:18:38 <ahills> is alpine your full-time job?
2015-06-05 13:18:51 <ncopa> part of it yes
2015-06-05 13:18:56 <ahills> jealous
2015-06-05 13:19:04 <ncopa> ;)
2015-06-05 13:46:03 <ncopa> ok
2015-06-05 13:46:06 <ncopa> here comes abuild
2015-06-05 13:46:28 <fabled> oh my. gcc 5.1
2015-06-05 13:46:38 <ncopa> hang on...
2015-06-05 13:46:39 <ncopa> soon
2015-06-05 13:47:26 <ncopa> here comes!
2015-06-05 13:47:29 <ncopa> whee
2015-06-05 13:48:34 <fabled> kaboom
2015-06-05 13:52:03 <ncopa> poor arm builder
2015-06-05 13:52:11 <ncopa> it was just done and then came gcc...
2015-06-05 13:52:12 <ncopa> :)
2015-06-05 14:09:11 <ncopa> whee
2015-06-05 14:09:20 <fabled> gcc-5 works
2015-06-05 14:09:26 <fabled> ... to show version atleast
2015-06-05 14:09:28 <fabled> :)
2015-06-05 14:25:50 <ncopa> whhe
2015-06-05 14:26:48 <fabled> that actually helps a lot when -dev does not depend on -doc
2015-06-05 14:27:06 <fabled> After this operation, 10 MiB of disk space will be freed.
2015-06-05 14:27:08 <fabled> :)
2015-06-05 14:31:43 <ncopa> 10MB saved only for those 10-20 packages?
2015-06-05 14:31:45 <ncopa> cool
2015-06-05 14:31:55 <ncopa> i think the docker ppl will be happy
2015-06-05 14:32:15 <fabled> ncopa, yes the -dev packages depended on big -doc and also perl-dev on miniperl
2015-06-05 14:32:27 <ncopa> cool
2015-06-05 14:33:04 <ncopa> we might manuall add some deps
2015-06-05 14:33:06 <ncopa> liek perl-dev
2015-06-05 14:33:13 <ncopa> you will always want 'perl' itself
2015-06-05 14:33:19 <ncopa> but its ont pulled in automatically
2015-06-05 14:33:35 <ncopa> flex-dev too
2015-06-05 14:33:41 <ncopa> probably python-dev too
2015-06-05 14:36:27 <[xming]> perl causes brain damage
2015-06-05 14:38:03 <ncopa> perl is a write-only language
2015-06-05 20:20:05 <bmurphy1_> I'm having trouble accessing a directory owned by a specific user as that user, I'm clearly missing something fundamental but I'm not sure what
2015-06-05 20:20:16 <bmurphy1_> here's what I'm seeing: http://sprunge.us/RCei
2015-06-05 20:20:35 <bmurphy1_> the nginx user can't access /var/lib/nginx even though it's 700 and owned by the nginx user
2015-06-06 09:41:03 <frol_> Hi guys! Could you please package Haskell? https://bugs.alpinelinux.org/issues/4257
2015-06-06 09:41:03 <frol_> It is known to compile for musl libc: http://www.reddit.com/r/haskell/comments/37m7q7/ghc_musl_easier_static_linking/
2015-06-06 09:41:03 <frol_> But because haskell compilation requires haskell, it is hard to setup. It will require 2 pull requests if I decide to contibute this package, but you, guys, can handle it better, I believe
2015-06-06 09:46:19 <barthalion> well, it would require some core-core dev assistance, I have no access to build servers to properly bootstrap it
2015-06-06 10:06:14 <frol_> BTW, is there any schedule for applying patchwork patches? http://patchwork.alpinelinux.org/project/aports/list/
2015-06-06 13:16:46 <ahills> <@ncopa> perl is a write-only language
2015-06-06 13:16:51 <ahills> truer words were never spoken
2015-06-06 13:39:23 <ahills> heh, shouldn't have updated... I can't get new terminals now
2015-06-06 13:39:33 <ahills> $ st
2015-06-06 13:39:33 <ahills> XOpenIM failed. Could not open input device.
2015-06-06 16:19:23 <kaniini> barthalion: i don't think it is required.  you could have two APKBUILDs, one which is ghc-bootstrap and the other that build-depends on ghc-bootstrap.
2015-06-06 16:20:49 <kaniini> initial bringup of ghc is a pain on new architectures though, so haskell should not use arch="all" unless an actual team comes together to maintain it
2015-06-06 16:26:48 <kaniini> frol__: i am build-testing your package now and will commit it as soon as its done
2015-06-06 16:26:49 <ahills> working on ghc now, actually
2015-06-06 16:27:05 <frol__> WOW!
2015-06-06 16:27:14 <frol__> That is awesome!
2015-06-06 16:27:18 <frol__> Thank you guys!
2015-06-06 16:27:30 <kaniini> ahills: we cannot use those binaries.
2015-06-06 16:27:57 <ahills> kaniini: yeah, I'm working on actually porting it
2015-06-06 16:28:07 <kaniini> ahills: however, the process could be used to generate our own bootstrap binaries
2015-06-06 16:28:15 <kaniini> it needs to be documented so its reproducable
2015-06-06 16:28:29 <ahills> I'm hoping to make a ghc-bootstrap package
2015-06-06 16:28:33 <ahills> that ghc can depend on
2015-06-06 16:28:44 <ahills> or just make one ghc package that bootstraps, then future versions don't
2015-06-06 16:28:51 <ahills> out to lunch, bbl
2015-06-06 16:33:19 <kaniini> frol__: if you want to verify the package works i'll move it to main
2015-06-06 16:34:23 <kaniini> ahills: http://patchwork.alpinelinux.org/patch/186/ NAK
2015-06-06 16:34:36 <kaniini> ahills: changing a library's ABI is something that we don't want to do
2015-06-06 16:35:43 <frol__> kaniini: I can test it from testing, no problem. Is it already there?
2015-06-06 16:35:58 <kaniini> yes
2015-06-06 16:36:02 <kaniini> apk add iojs@testing
2015-06-06 16:36:41 <kaniini> http://patchwork.alpinelinux.org/patch/187/ why do we need libbsd-dev 0.7
2015-06-06 16:37:26 <ahills> kaniini: for fortune :)
2015-06-06 16:37:36 <kaniini> no, why does fortune require it.
2015-06-06 16:37:55 <kaniini> it builds fine with current libbsd 0.6
2015-06-06 16:38:07 <ahills> If there's a Linux port it might not, I was porting openbsd's version
2015-06-06 16:38:15 <ahills> Where are you getting your fortune?
2015-06-06 16:38:24 <kaniini> your package
2015-06-06 16:38:28 <kaniini> builds fine with libbsd 0.6
2015-06-06 16:38:34 <ahills> It didn't build for me...
2015-06-06 16:38:44 <kaniini> did for me
2015-06-06 16:39:03 <ahills> Weird, I'll try again
2015-06-06 16:59:16 <frol__> kaniini: it works! (I had to upgrade libstdc++ to 5.1 from edge, though)
2015-06-06 17:10:33 <kaniini> frol__: do you wish to be maintainer of it
2015-06-06 17:11:53 <frol__> kaniini: not really, I'm a Python dev mostly and I'm just experimenting with JS now, so I won't be able to monitor new versions, unfortunatelly :(
2015-06-06 17:19:52 <kaniini> ok
2015-06-06 17:20:00 <kaniini> should probably keep it in testing then
2015-06-06 18:05:27 <ahills> kaniini: someone must have slipped something in my drink when I was building fortune earlier... I honestly have no idea what I was thinking
2015-06-06 18:11:05 <pnutzh4x0r> hmm, i guess http://patchwork.alpinelinux.org/patch/184/ can be closed/removed
2015-06-06 18:31:02 <ahills> kaniini: it seems that a chroot with glibc is the only way to produce a cross-compiler for ghc to bootstrap the build
2015-06-06 18:31:17 <ahills> I'm not really comfortable building that sort of environment in an APKBUILD...
2015-06-06 18:31:18 <kaniini> ahills: correct
2015-06-06 18:31:34 <ahills> it requires root privileges
2015-06-06 18:31:35 <kaniini> ahills: what i mean is that we need scripts to build the bootstrap binaries
2015-06-06 18:31:47 <ahills> hmm, ok
2015-06-06 18:31:51 <kaniini> ahills: it can happen separate from APKBUILD, we just can't use some 3rd-party binaries
2015-06-06 18:32:02 <ahills> kaniini: scripts that set up a container would be acceptable?
2015-06-06 18:32:16 <kaniini> ahills: sure, and then scripts to build the binaries
2015-06-06 18:32:24 <kaniini> ahills: and then we have an APKBUILD which unpacks them
2015-06-06 18:32:31 <ahills> ok
2015-06-06 18:32:42 <ahills> after the first time, it can depend on itself
2015-06-06 18:33:16 <kaniini> yes but we need the scripts for new archs
2015-06-06 18:33:18 <kaniini> :)
2015-06-06 18:33:23 <ahills> of course!
2015-06-06 18:33:29 <ahills> I just mean for the build procedure
2015-06-06 18:33:40 <ahills> I also can't really test this on x86
2015-06-06 18:33:54 <ahills> and I don't want to dedicate my rpi to building ghc on arm for the next six months :P
2015-06-06 18:34:17 <kaniini> i believe we have a quad-core exynos odroid box for porting
2015-06-06 19:12:14 <ahills> kaniini: unfortunately I need to postpone my ghc activities, please don't expect results today
2015-06-06 19:12:26 <kaniini> 
ACTION wasn't expecting results ;)

2015-06-06 19:16:09 <ahills> !!!
2015-06-07 21:21:33 <Guest70161> Install freeradius leads to unsatisfied constraints between freeradius-lib and freeradius
2015-06-08 06:37:28 <ncopa> rnalrd: freeradius/freeradius3 is broke
2015-06-08 06:37:48 <ncopa> i think it might be due a to an abuild issue
2015-06-08 06:38:05 <ncopa> i think provides="freeradius3" gets into all subpackages
2015-06-08 06:38:06 <rnalrd> uh
2015-06-08 06:38:21 <rnalrd> i didn't add that
2015-06-08 06:38:46 <rnalrd> was is it done to prevent broken upgrade to 3.2?
2015-06-08 06:39:11 <rnalrd> s/is//
2015-06-08 06:39:18 <ncopa> yea
2015-06-08 06:39:26 <ncopa> i htink it was so help upgrade to v3.2
2015-06-08 06:39:33 <ncopa> jbilyk
2015-06-08 06:39:48 <ncopa> http://git.alpinelinux.org/cgit/aports/commit/main/freeradius?id=1022340ce2279c7f076744df1ae8993dd3f0051d
2015-06-08 06:41:02 <rnalrd> i think we should remove it and let users manually handle it
2015-06-08 06:41:12 <rnalrd> we have that in release notes
2015-06-08 06:41:23 <ncopa> i think we can try make it work
2015-06-08 06:41:28 <ncopa> i dont think its difficult
2015-06-08 06:41:33 <ncopa> and we can remove it for v3.3
2015-06-08 06:41:42 <ncopa> i did the same with mysql-dev
2015-06-08 06:41:54 <ncopa> hm
2015-06-08 07:14:47 <ScrumpyJack> morning
2015-06-08 07:20:24 <ncopa> hi
2015-06-08 07:20:25 <ncopa> hum
2015-06-08 07:20:36 <ncopa> provides is nasty
2015-06-08 07:21:05 <ncopa> i think this is a real problem
2015-06-08 07:21:10 <ncopa> in the ABPKUILD format
2015-06-08 07:22:02 <ncopa> if you set provides="something" in the global scope, then will all your subpackages have the same provides
2015-06-08 07:22:19 <ncopa> which probably is not what you want
2015-06-08 07:22:23 <ncopa> same with depends
2015-06-08 07:22:42 <ncopa> and same with install_if
2015-06-08 07:23:45 <ncopa> so i think we need reset those for subpackages
2015-06-08 07:23:46 <ncopa> but
2015-06-08 07:23:51 <ncopa> you do want pkgver
2015-06-08 07:37:50 <ncopa> rnalrd: i'm about to do this with freeradius: http://sprunge.us/iMeN
2015-06-08 07:38:02 <ncopa> note that i remove freeradius-lib from depends
2015-06-08 07:38:09 <ncopa> and add ldpath=...
2015-06-08 07:38:26 <ncopa> this will make abuild figure out the .so depends himself
2015-06-08 07:38:41 <ncopa> ok?
2015-06-08 07:39:01 <rnalrd> sounds good :)
2015-06-08 07:39:05 <ncopa> thanks
2015-06-08 07:44:43 <ncopa> rnalrd: maybe move pmacct back to main too ;)
2015-06-08 07:44:57 <rnalrd> yeah, wanna test it first
2015-06-08 07:45:02 <ncopa> testing then
2015-06-08 07:45:09 <ncopa> builders does not build anything in unmaintained
2015-06-08 07:47:05 <rnalrd> i know
2015-06-08 07:47:12 <rnalrd> it starts with default config
2015-06-08 07:47:14 <rnalrd> so moving to main
2015-06-08 10:02:58 <ScrumpyJack> so i want to make a deployable image of alpine linux
2015-06-08 10:04:15 <ScrumpyJack> ideally i'd be able to install to HDD, set up packages and configure the system to behave exactly as I want it
2015-06-08 10:06:45 <ScrumpyJack> short of using something like clonezilla, is there a neat way of cloning this installation to multiple pieces of (identical) hardware?
2015-06-08 10:08:22 <ScrumpyJack> I can dd over a netcat, but i was wondering if there was a simpler way with a USB
2015-06-08 10:09:11 <ScrumpyJack> perhaps then i could run setup-alpine and would end up with a clone
2015-06-08 10:14:14 <fabled> ScrumpyJack, is it enough if you create answer file with "setup-alpine -c" and use that?
2015-06-08 10:20:58 <fcolista> should is somewhere in b.a.o. the request for a kickstart file integration
2015-06-08 10:21:04 <fcolista> *should be
2015-06-08 10:57:57 <ScrumpyJack> hmm, yes i see what you mean. it's not exaclt what I want. the answer file is good for setup-alpine, but i'm looking for a simple way of clone an HDD install on multiples of the same hardware, without building a custom ISO
2015-06-08 10:58:44 <ScrumpyJack> once clone, I can run setup-alpine -f hostname.answerfile or something
2015-06-08 11:00:21 <ScrumpyJack> or get what i've built on one HDD install to many
2015-06-08 11:08:58 <bCQc0P3DIqbh> if you run setup-alpine from an already set up system it copies over all installed packages iirc
2015-06-08 11:39:09 <ahills> ncopa: does the builder cross-compile for x86 or build on an x86 platform?
2015-06-08 11:41:52 <ncopa> i build on x86 platform
2015-06-08 11:42:03 <ncopa> in a lxc container to be exact
2015-06-08 11:43:31 <ahills> hmm, it's easy to run an x86 container on an x86_64 system?
2015-06-08 11:43:39 <ncopa> yes
2015-06-08 11:43:44 <ahills> I haven't tested any of my packages on x86, time to change that
2015-06-08 11:44:43 <ncopa>  lxc-create -t alpine -n containername -- --arch=x86
2015-06-08 11:49:04 <vkrishn> if looking for simpler syntax for configs, http://ne-on.org/
2015-06-08 11:50:00 <ahills> vkrishn: looks identical to yaml
2015-06-08 11:50:44 <vkrishn> yes, the page states some difference at bottom
2015-06-08 11:54:41 <barthalion> yes, this is what we need
2015-06-08 11:54:53 <barthalion> yet another markup language
2015-06-08 11:55:21 <barthalion> vkrishn, nobody is going to use it, yaml and json are more or less considered standard
2015-06-08 11:56:11 <ahills> if json allowed trailing commas in lists, it would probably be my favorite
2015-06-08 11:59:22 <vkrishn> then maybe stick to json, as its making lots of rfc, http://tools.ietf.org/html/new-rfcs.rss
2015-06-08 12:00:30 <ahills> not my favorite for configurations (although it's not too bad to work with awall)
2015-06-08 12:00:40 <ahills> it's just a nice data transport because it's human-parseable
2015-06-08 12:01:39 <ahills> I'd rather not need to type so many quote characters when writing a configuration, I prefer kaniini's format if a full tree is needed, or the simple sh-style
2015-06-08 12:06:50 <vkrishn> what plans for AL 3.3 init system ?
2015-06-08 12:07:00 <vkrishn> openrc, runit, s6 !
2015-06-08 12:08:06 <barthalion> none at the moment
2015-06-08 12:11:08 <ahills> at some point I will try runit with alpine, but I think the work to convert from openrc is pretty severe
2015-06-08 12:21:56 <ScrumpyJack> hmm, how come search in resolv.conf isn't being appended to my hostname queries?
2015-06-08 12:24:18 <ScrumpyJack> so in /etc/resolv.conf i have search b.c
2015-06-08 12:24:45 <ScrumpyJack> I'd expect ping a to resolve a.b.c
2015-06-08 12:31:50 <ScrumpyJack> arg! "After some discussion, search domains were deemed a largely-harmful feature, so support for them was not added."
2015-06-08 12:54:53 <ahills> ScrumpyJack: link?
2015-06-08 12:55:06 <ahills> is that musl?
2015-06-08 12:55:21 <ahills> I am very interested in having that feature...
2015-06-08 13:01:11 <ncopa> we should talk about it in #alpine-linux where dalias (musl dev) hangs around
2015-06-08 13:33:29 <ahills> ncopa: the autobuilder repository is software for the builders to respond to algitbot?
2015-06-08 13:34:31 <ncopa> yes, deprecated
2015-06-08 13:35:11 <ahills> ah, ok
2015-06-08 13:35:51 <ahills> is there a way to configure abuild to automatically execute a command after a successful complete build, e.g. to rsync packages somewhere?
2015-06-08 13:38:17 <barthalion> &&
2015-06-08 13:38:18 <barthalion> :P
2015-06-08 13:38:41 <ahills> haha
2015-06-08 13:38:59 <ahills> I guess I could inotify on ~/packages
2015-06-08 13:56:02 <ncopa> inotty on ~/packages sounds cool :)
2015-06-08 13:56:41 <ncopa> the current build servers use aports-build
2015-06-08 13:57:12 <ncopa> which agains uses buildrepo
2015-06-08 13:58:12 <ahills> I want successful builds on my laptop's build container to push my new stuff to my build vm and build it there
2015-06-08 13:58:33 <ahills> I should probably just use git hooks
2015-06-08 13:59:54 <ahills> I also want it to ping me in irc when it finishes
2015-06-08 14:02:19 <fcolista> fabled, remember me: now that alpine 3.2 is shipped, i'll try to boot a beaglebone with that (that's not beaglbone black). Once i download the arm image, i only need to have the sd bootable, unpack the arm image and copy the MLO with u-boot.img on the root.
2015-06-08 14:02:30 <fcolista> Is that enough?
2015-06-08 14:05:10 <fabled> i've never tried beagle bone. but i think the kernel + u-boot should be enabled.
2015-06-08 14:05:12 <fabled> so yes
2015-06-08 14:05:15 <fabled> well
2015-06-08 14:05:31 <fabled> check the README, i'm not fully sure how the sd/mmc needs to be prepared
2015-06-08 14:05:56 <fabled> http://archlinuxarm.org/platforms/armv7/ti/beaglebone --> installation
2015-06-08 14:05:58 <fcolista> about beaglebone it says to copy MLO and uboot on the root
2015-06-08 14:06:17 <fcolista> yes, i'm there
2015-06-08 14:06:27 <fabled> it uses dd
2015-06-08 14:06:33 <fabled> seems my README says wrong
2015-06-08 14:06:45 <fcolista> it was my doubt.
2015-06-08 14:06:53 <fcolista> README also says dd for rpi
2015-06-08 14:06:58 <fcolista> but not for beagle
2015-06-08 14:07:03 <fabled> ?
2015-06-08 14:07:08 <fabled> no dd is for wand
2015-06-08 14:07:13 <fcolista> oh right
2015-06-08 14:07:20 <fcolista> bad memory
2015-06-08 14:07:34 <fabled> dd if=mnt/bootMLO of=/dev/sdX count=1 seek=1 conv=notrunc bs=128k
2015-06-08 14:07:34 <fabled> dd if=u-boot.img of=/dev/sdX count=2 seek=1 conv=notrunc bs=384k
2015-06-08 14:07:52 <fcolista> is it fine to use ext4?
2015-06-08 14:07:56 <fcolista> Or i should use vfat?
2015-06-08 14:08:08 <fabled> mmm
2015-06-08 14:08:24 <ahills> vfat is always the safest...
2015-06-08 14:08:26 <fabled> i think either works, but you might need to tune the alpine boot command
2015-06-08 14:08:30 <fabled> for ext4
2015-06-08 14:08:33 <fabled> so try first vfat
2015-06-08 14:20:27 <fcolista> no way to boot so far
2015-06-08 14:20:45 <fcolista> user leds ar all powered on
2015-06-08 14:20:50 <fabled> i'm off
2015-06-08 14:20:54 <fabled> hope you get it work
2015-06-08 14:21:00 <fcolista> well 3/4.  C u fabled
2015-06-08 14:22:56 <ahills> ncopa: does aports-build live in aports?
2015-06-08 14:39:24 <ncopa> ahills: yes
2015-06-08 14:46:45 <ahills> what's the preferred method of fetching/building from git when no tarballs are available? I'm guessing http://wiki.alpinelinux.org/wiki/APKBUILD_examples:Git_checkout isn't really relevant for me
2015-06-08 14:49:14 <ncopa> i suppose creating a snaphot tarball
2015-06-08 14:49:30 <ncopa> but what i normally do is ask maintainer to make release tarball
2015-06-08 14:49:51 <ncopa> https://github.com/Cloudef/xcmenu/issues/12
2015-06-08 14:51:06 <ahills> good idea
2015-06-08 14:52:21 <ahills> does github automatically create releases for v* tags?
2015-06-08 15:03:43 <ncopa> ahills: yes, the automatically create archivs for v[0-9]* tagz
2015-06-08 15:03:49 <ncopa> tags*
2015-06-08 15:04:41 <tdtrask> :(
2015-06-08 15:05:12 <tdtrask> ncopa: x86 worked but x86_64 failed?
2015-06-08 15:05:25 <ncopa> http://bld2.alpinelinux.org/buildlogs/build-edge-x86_64/testing/acf-provisioning-grandstream-ata/acf-provisioning-grandstream-ata-1.0.7.3-r0.log
2015-06-08 15:05:31 <ncopa> apparently so
2015-06-08 15:05:53 <ncopa> curl: (18) transfer closed with 569555 bytes remaining to read
2015-06-08 15:06:01 <tdtrask> how to get it to retry?
2015-06-08 15:06:51 <ahills> build64-2-5: retry
2015-06-08 15:07:14 <tdtrask> ncopa: thanks
2015-06-08 15:07:22 <ncopa> mosquitto_pub -h msg.alpinelinux.org -t 'git/aports/master' -m retry
2015-06-08 15:07:26 <ncopa> is what i did
2015-06-08 15:07:31 <ahills> can anyone publish?
2015-06-08 15:07:35 <ncopa> yup
2015-06-08 15:07:56 <ahills> will you wait for abuse before you secure it?
2015-06-08 15:08:04 <ncopa> probably
2015-06-08 15:08:28 <ncopa> we used to use IRC as message broker before
2015-06-08 15:08:33 <ahills> yes I saw
2015-06-08 15:08:42 <ahills> from the deprecated scripts I was reading
2015-06-08 15:08:49 <ncopa> but we can kick things on irc
2015-06-08 15:08:58 <ncopa> and we see it pretty soon
2015-06-08 15:09:02 <ncopa> thats not so with mqtt
2015-06-08 15:09:05 <ncopa> well
2015-06-08 15:09:05 <ahills> if you're awake
2015-06-08 15:09:08 <ncopa> yes
2015-06-08 15:09:27 <ncopa> might be an idea to secure it with password
2015-06-08 15:09:36 <ncopa> even if its a weak password
2015-06-08 15:09:45 <ahills> mqtt supports very basic auth I think
2015-06-08 15:09:48 <ncopa> yes
2015-06-08 15:10:07 <ncopa> that is one of the reasons i picked mqtt
2015-06-08 15:10:18 <ncopa> it had the feature if ever needed
2015-06-08 15:10:44 <ncopa> ok i better go home
2015-06-08 15:10:51 <ahills> goodnight
2015-06-08 15:10:54 <ncopa> see u
2015-06-08 15:50:21 <ahills> thank you SO MUCH to whomever made update_config_sub
2015-06-08 15:50:36 <ahills> even a little less pain from autotools can make the day brighter
2015-06-09 11:19:47 <ncopa> re terminus-font
2015-06-09 11:19:57 <ncopa> do we want call it ttf-terminus-font?
2015-06-09 11:20:14 <ncopa> we have lots other ttf-* packages for fonts
2015-06-09 11:20:31 <ncopa>  apk search --exact ttf-*
2015-06-09 11:21:28 <vkrishn> ncopa, zeromq v4.1.1 is a stable release
2015-06-09 11:21:28 <ncopa> ok
2015-06-09 11:21:34 <ncopa> seems to not be ttf font
2015-06-09 11:22:30 <vkrishn> also php-phalcon bump to v2.0.2 pls
2015-06-09 11:22:47 <ncopa> zeromq depends on libsodium
2015-06-09 11:22:50 <ncopa> which is in testing
2015-06-09 11:23:13 <vkrishn> ah
2015-06-09 11:26:46 <ncopa> do you need PGM extension?
2015-06-09 11:26:51 <ncopa> its default off
2015-06-09 11:28:11 <vkrishn> its nice feature, multicast, iirc
2015-06-09 11:28:28 <ncopa> yes but do we need it?
2015-06-09 11:28:41 <vkrishn> not use uptil now
2015-06-09 11:28:48 <vkrishn> used*
2015-06-09 11:29:41 <vkrishn> almost set aside nanomsg/zeromq after mosquitto
2015-06-09 11:42:45 <vkrishn> zeromq-4.1.1 does not have foreign folder , but zeromq-4.0.6 does
2015-06-09 11:43:31 <vkrishn> foreign/openpgm/libpgm-5.2.122~dfsg.tar.gz
2015-06-09 11:52:11 <vkrishn> or maybe have libpgm https://code.google.com/p/openpgm/ separately
2015-06-09 12:06:45 <ncopa> vkrishn: yes, i suppose we'd need use openpgm if we need that
2015-06-09 12:06:54 <ncopa> i disable pgm since nobody uses it
2015-06-09 12:06:59 <ncopa> unused features = bloat
2015-06-09 12:07:27 <ncopa> i'll add it back if someone asks for it
2015-06-09 12:08:47 <vkrishn> ok, thanks
2015-06-09 12:14:08 <ncopa> now what...
2015-06-09 12:20:42 <vkrishn> oh, main/openpgm/APKBUILD its there
2015-06-09 12:57:30 <ncopa> i have the first c++ rebuild issue with gcc5
2015-06-09 12:57:35 <ncopa> protobuf needs rebuild i think
2015-06-09 13:06:29 <fcolista> umh: /bin/ash: svnversion: not found
2015-06-09 13:08:25 <fcolista> I don't have subversion but here builds
2015-06-09 13:11:32 <barthalion> which package?
2015-06-09 13:11:49 <fcolista> nzbget fails to build 'cause does not find subversion
2015-06-09 13:11:59 <fcolista> http://bld2.alpinelinux.org/buildlogs/build-edge-x86_64/testing/nzbget/nzbget-15.0-r0.log
2015-06-09 13:12:28 <fcolista> It's calledi in Makefile
2015-06-09 13:12:30 <fcolista> *called
2015-06-09 13:12:32 <barthalion> it's not the main problem I guess
2015-06-09 13:12:40 <barthalion> File exists is bigger one
2015-06-09 13:12:41 <ncopa> looks like its parallel install issue
2015-06-09 13:12:54 <barthalion> you should do make install -j1 or fix Makefile to use mkdir -p
2015-06-09 13:13:10 <fcolista> again parallel build issue.
2015-06-09 13:13:57 <ahills> Why are so many people bad at designing build systems?
2015-06-09 13:14:25 <barthalion> why are so many people bad at writing correct Makefiles?
2015-06-09 13:14:31 <barthalion> ftfy
2015-06-09 13:15:31 <fcolista> what do you think about %182
2015-06-09 13:15:32 <fcolista> ?
2015-06-09 13:15:45 <fcolista> main/nginx-lua
2015-06-09 13:16:07 <ncopa> pkgrel should be 0
2015-06-09 13:16:32 <fcolista> yes, it's the arch way which starts from 1
2015-06-09 13:17:11 <ncopa> not using luajit sounds good to me
2015-06-09 13:17:30 <ncopa> i prefer normal lua so we dont need to paxmark
2015-06-09 13:17:48 <ncopa> otherwise, i think it looks okish
2015-06-09 13:18:01 <ncopa> i am not too happy with the general nginx situation
2015-06-09 13:18:19 <ncopa> that we need have different nginx packages for every single module
2015-06-09 13:18:19 <fcolista> yes, remember that
2015-06-09 13:18:24 <fcolista> with modules
2015-06-09 13:18:29 <ncopa> so
2015-06-09 13:18:30 <barthalion> domo arigato upstream
2015-06-09 13:18:34 <barthalion> well, no so
2015-06-09 13:18:37 <barthalion> no so at all
2015-06-09 13:18:46 <ncopa> i complained upstream
2015-06-09 13:18:51 <ncopa> and i think they have it on their todo
2015-06-09 13:18:55 <ncopa> to modularize
2015-06-09 13:19:09 <barthalion> (y)
2015-06-09 13:19:28 <ncopa> the current problem is
2015-06-09 13:19:32 <ncopa> he fixed the init.d script
2015-06-09 13:19:37 <ncopa> but there are other nginx package
2015-06-09 13:19:44 <ncopa> with other init.d script
2015-06-09 13:19:48 <ncopa> which works differently
2015-06-09 13:20:10 <ncopa> so we might want ship a separate nginx-initscripts
2015-06-09 13:20:28 <ncopa> and have nginx/nginx-lua/nginx-othermodule to depend on that
2015-06-09 13:20:38 <ncopa> so we dont need fix 3 different init.d scripts
2015-06-09 13:22:47 <ncopa> i think libbsd update was NACKed by kaniini
2015-06-09 13:23:46 <fcolista> stupid question for git. I git-am the nginx-lua, then reset pkgrel to 0. Is this correct? Or builder will create first a -r1 and then a -r0 ?
2015-06-09 13:24:31 <fcolista> How do i modify an already committed modify ? git commit -v --amend is the way?
2015-06-09 13:24:52 <fcolista> So i'll directly set pkgrel to 0, without a doulbe commit.
2015-06-09 13:24:56 <fcolista> *double.
2015-06-09 13:25:56 <ahills> ncopa: it was, for some reason fortune wasn't building for me with 0.6, but it seems it's fine
2015-06-09 13:26:37 <ahills> fcolista: git commit, git rebase -i HEAD^^
2015-06-09 13:28:55 <fcolista> thx ahills
2015-06-09 13:30:08 <ahills> :)
2015-06-09 13:35:51 <ahills> What do you guys use for PDF?
2015-06-09 13:36:01 <ahills> Reading, I mean
2015-06-09 13:39:02 <clandmeter> firefox :)
2015-06-09 13:39:37 <kaniini> ncopa: correct, it made changes to the ABI which were different than upstream
2015-06-09 13:40:34 <ncopa> fcolista: if you add a pkgrel=0 as incremental commit before you git push, the will builder never build the pkgrel=1 version
2015-06-09 13:40:47 <fcolista> aah
2015-06-09 13:40:51 <fcolista> ok
2015-06-09 13:40:55 <ncopa> its a bit ugly though, so you might want do --amend
2015-06-09 13:41:01 <ncopa> but the will pwclient not pick it up
2015-06-09 13:41:13 <ncopa> and the author will get conflict
2015-06-09 13:41:16 <fcolista> --amend is not only related to message then
2015-06-09 13:41:18 <ncopa> which i think is ok...
2015-06-09 13:41:23 <ncopa> correct
2015-06-09 13:41:37 <ncopa> you can change pkgrel=0, git add APKBUILD; git commit --amend
2015-06-09 13:41:45 <ncopa> and the original author is kept
2015-06-09 13:41:50 <ncopa> but with your fix
2015-06-09 13:41:52 <barthalion> patchwork dislikes it
2015-06-09 13:42:13 <ncopa> you can manually mark the patch 'superseded' or similar
2015-06-09 13:43:31 <ncopa> seems like that is exactly what happened with fortune
2015-06-09 13:43:59 <ncopa> commit b0ac98ed1cabfb4add39d1ee184e25e34e872e93
2015-06-09 13:43:59 <ncopa> Author: Andrew Hills <ahills@ednos.net>
2015-06-09 13:43:59 <ncopa> Commit: William Pitcock <nenolod@dereferenced.org>
2015-06-09 13:43:59 <ncopa>     testing/fortune: new aport
2015-06-09 13:44:26 <ncopa> patchwork didnt pick it up: http://patchwork.alpinelinux.org/patch/187/
2015-06-09 13:44:45 <ncopa> so i manually have to mark it
2015-06-09 13:44:51 <ncopa> i think thats ok
2015-06-09 13:45:10 <clandmeter> patchwork checks the hash of the commit, so if you change it it wont find it.
2015-06-09 13:45:15 <ncopa> yeah
2015-06-09 13:45:23 <ncopa> which is correct
2015-06-09 13:45:29 <ncopa> it is another commit, another patch
2015-06-09 13:52:26 <ncopa> linux-headers?
2015-06-09 13:52:34 <fabled> no, armhf only
2015-06-09 13:52:45 <clandmeter> rpi only
2015-06-09 13:53:05 <ncopa> fabled: while at rpi, could you please look at http://patchwork.alpinelinux.org/patch/106/
2015-06-09 13:53:16 <ncopa> oh
2015-06-09 13:53:20 <ncopa> you did comment it already
2015-06-09 13:53:22 <fabled> yes
2015-06-09 13:53:26 <fabled> cc'd you too
2015-06-09 13:53:43 <ncopa> it probably drowned in my inbox
2015-06-09 14:20:44 <ncopa> thanks for the patches you send ppl
2015-06-09 14:23:23 <ncopa> bah fcolista beat me to it :)
2015-06-09 14:23:31 <fcolista> :)
2015-06-09 14:23:41 <ncopa> the kamailio patches does not apply
2015-06-09 14:24:10 <ncopa> clandmeter: thanks!
2015-06-09 14:24:18 <clandmeter> its just a start
2015-06-09 14:24:19 <kaniini> noooo
2015-06-09 14:24:23 <clandmeter> :)
2015-06-09 14:24:24 <ncopa> please remove it from unmaintained too
2015-06-09 14:24:27 <kaniini> boooo
2015-06-09 14:24:32 <ncopa> :D
2015-06-09 14:24:38 <kaniini> will we be adding systemd next
2015-06-09 14:24:59 <ncopa> kaniini: not til they support musl libc
2015-06-09 14:25:12 <kaniini> i had a bad experience with pulseaudio once
2015-06-09 14:25:20 <kaniini> i plugged some headphones in
2015-06-09 14:25:26 <kaniini> played an mp3 file
2015-06-09 14:25:31 <kaniini> and it was at 100% volume
2015-06-09 14:25:35 <ncopa> aw
2015-06-09 14:25:49 <kaniini> then i was told it was by design
2015-06-09 14:25:58 <clandmeter> i have that with mpd all the time without PA :)
2015-06-09 14:26:28 <kaniini> for my intel hda card, ALSA keeps the headphone jack muted until i explicitly unmute it
2015-06-09 14:32:58 <jomat> ncopa: thx for applying the patch. This should have been the correct one: https://0.jmt.gr/?773b82c00e98d452#cKfG3hb109sFowxvxpiG+hEaL0jc8GZH65UuRwN7GaM= ... well, too late now :-)
2015-06-09 14:33:12 <jomat> I sent it to the wrong email adress...
2015-06-09 14:34:25 <ncopa> jomat: yeah , i fixed it manually
2015-06-09 14:34:26 <ncopa> thanks
2015-06-09 14:34:33 <ncopa> and i added a few || return 1
2015-06-09 14:34:35 <ncopa> while at it
2015-06-09 14:34:46 <ncopa> so we catch future build failures
2015-06-09 14:37:12 <jomat> Yeah, I know about them... I don't know if you've read my question regarding set -x somehow in APKBUILDs, didn't check backlog too regulary, I'm quite busy these days
2015-06-09 14:37:37 <jomat> ehrr... set -e of course
2015-06-09 18:01:24 <shafire> hi
2015-06-09 18:01:33 <shafire> is openstack for alpine possible?
2015-06-09 18:01:46 <barthalion> sure
2015-06-09 18:01:53 <barthalion> nothing impossible about it
2015-06-09 18:02:14 <shafire> because all my tickets are closed because of glibc dependency :D
2015-06-09 18:02:22 <barthalion> huh
2015-06-09 18:02:30 <shafire> my last*
2015-06-09 18:02:32 <kunkku> openstack is python
2015-06-09 18:02:35 <shafire> not specific about openstack
2015-06-09 18:02:40 <barthalion> which ticket exactly?
2015-06-09 18:02:47 <shafire> sorry, you understood me wrong
2015-06-09 18:02:57 <barthalion> ok, I see now
2015-06-09 18:02:58 <shafire> there is no ticket for openstack
2015-06-09 18:03:05 <barthalion> openstack is written in python, so it's not a big problem
2015-06-09 18:03:16 <shafire> do you use openstack?
2015-06-09 18:03:18 <barthalion> except it's literally shitload of packages, if you want to have full stack
2015-06-09 18:03:32 <kunkku> packaging it is a lot of work
2015-06-09 18:03:32 <barthalion> yes, more or less I'm paid for working on openstack
2015-06-09 18:03:35 <shafire> I saw the github project
2015-06-09 18:04:05 <shafire> barthalion: do you like it?
2015-06-09 18:04:16 <barthalion> and even if amount of packages to make doesn't scare you, you need to track quite explicit dependencies in global-requirements.txt
2015-06-09 18:04:48 <barthalion> it's hard to say if I like it, but it's really amazing here and there
2015-06-09 18:04:52 <barthalion> that it actually works
2015-06-09 18:05:06 <shafire> I am using xen/kvm by hand, everything manual
2015-06-09 18:05:15 <shafire> I lose to much time :S need to automate stuff
2015-06-09 18:05:36 <barthalion> take a look at fuel</shameless plug>
2015-06-09 18:05:50 <barthalion> we're going to release 6.1 soon
2015-06-09 18:06:08 <kunkku> barthalion: you work for Mirantis?
2015-06-09 18:06:09 <barthalion> no idea what's more around… probably something from Red Hat
2015-06-09 18:06:13 <barthalion> kunkku: yes
2015-06-09 18:06:30 <barthalion> shafire: also probably Canonical with its juju allows to deploy openstack easily
2015-06-09 18:07:15 <shafire> barthalion: is this a russian company?
2015-06-09 18:07:22 <barthalion> not really
2015-06-09 18:07:31 <kunkku> the founders are
2015-06-09 18:07:45 <shafire> because the dns servers are all in russian :D
2015-06-09 18:08:33 <barthalion> but it's hard to hide that majority of employees are from UA and RU
2015-06-09 18:09:14 <shafire> I am curious if your office in kharkiv is still alive
2015-06-09 18:09:38 <barthalion> why wouldn't it be?
2015-06-09 18:10:08 <barthalion> the Lviv one is fine too
2015-06-09 18:10:16 <shafire> lviv is okay
2015-06-09 18:11:05 <shafire> okay
2015-06-09 18:11:31 <shafire> I checked, kharkiv is not in question
2015-06-09 18:11:48 <barthalion> what are you actually looking for?
2015-06-09 18:12:09 <shafire> nothing, all okay
2015-06-09 18:13:21 <barthalion> I really wish someone could pay me for packaging Openstack for alpine, but we're not here yet apparently :P
2015-06-09 18:13:48 <shafire> how much would you take?
2015-06-09 18:14:07 <shafire> 1 beer? :D
2015-06-09 18:15:42 <barthalion> can you survive a month with one beer? :p
2015-06-09 18:17:25 <shafire> :)
2015-06-09 18:22:08 <ahills> barthalion: fuel is in alpine?
2015-06-09 18:22:15 <ahills> and, does it come with a web interface?
2015-06-09 18:23:58 <barthalion> no, fuel uses centos for master node and can deploy ubuntu or centos; yes, it has web ui
2015-06-09 18:24:25 <barthalion> I guess almost every openstack deployment solution has it
2015-06-09 18:31:11 <ahills> ah, it's only for deploying ubuntu/centos?
2015-06-09 18:56:21 <barthalion> ahills: yeah
2015-06-10 04:27:54 <sugith> hi
2015-06-10 06:58:34 <coredumb> has there been any attempt to patch gradm to support symlinks on Alpine ?
2015-06-10 06:59:30 <coredumb> Using RBAC with the current gradm seems pretty useless on busybox to me
2015-06-10 07:16:25 <ncopa> coredumb: i've never used gradm feature
2015-06-10 07:36:31 <flatroze> Hello.
2015-06-10 07:44:29 <mosez> is it just my machine or is amavisd broken?
2015-06-10 07:44:48 <mosez> # /usr/sbin/amavisd -h
2015-06-10 07:44:48 <mosez> ERROR: MISSING REQUIRED BASIC MODULES: Net::LibIDN
2015-06-10 07:45:14 <coredumb> ncopa: That's very good at locking down completely the OS
2015-06-10 07:45:56 <coredumb> I bet a lot of grsec users use grsec only for RBAC ^^
2015-06-10 07:46:10 <coredumb> but it doesn't support symlinks
2015-06-10 07:46:24 <coredumb> well it actually translates them to the target
2015-06-10 07:46:36 <coredumb> which would be problematic on busybox
2015-06-10 07:46:41 <mosez> rnalrd: you are the maintainer, is it possible that you missed a dependency? :)
2015-06-10 07:46:53 <rnalrd> hi
2015-06-10 07:46:58 <rnalrd> could be
2015-06-10 07:47:09 <rnalrd> i'm not using it anymore actually
2015-06-10 07:47:22 <rnalrd> and I relied only on docs to find dependencies
2015-06-10 07:47:22 <mosez> there is no package for Net::LibIDN and it's failing to compile with cpan :(
2015-06-10 07:47:25 <rnalrd> let me add it
2015-06-10 07:48:08 <mosez> https://gist.github.com/tboerger/12a6fff214653f6d8708 this is a dependency list within the amavisd script
2015-06-10 07:48:19 <mosez> so what are you using for virus scanning? :)
2015-06-10 07:48:35 <rnalrd> clamsmtp
2015-06-10 07:50:41 <mosez> interesting
2015-06-10 07:51:04 <mosez> but as i want to use all features of a specific postfix webui i have to use amavisd-new -.-
2015-06-10 07:52:37 <mosez> rnalrd: awesome if you take a look at the dependencies. please send me a ping if you've got something new.
2015-06-10 07:53:53 <rnalrd> sure, i'm already on it
2015-06-10 07:53:59 <flatroze> I've noticed that you have 2014 as the year of copyright in the footer on the website, and 2013 on the forum; felt like it's worth reminding you to update it
2015-06-10 07:56:54 <ncopa> coredumb: i understand the problem with the symlinks
2015-06-10 07:57:06 <ncopa> i dont have any ideas for solutions
2015-06-10 07:57:33 <coredumb> I guess patching or harassing Brad
2015-06-10 07:57:35 <coredumb> :D
2015-06-10 07:59:37 <barthalion> ncopa: who's responsible for footers? :P
2015-06-10 08:00:23 <ncopa> barthalion: footers?
2015-06-10 08:00:37 <barthalion> 09:53:53 <flatroze> I've noticed that you have 2014 as the year of copyright in the footer on the website, and 2013 on the forum; felt like it's worth reminding you to update it
2015-06-10 08:01:04 <rnalrd> mosez ^^^^
2015-06-10 08:01:11 <ncopa> clandmeter: do you fix forums footer copyright year?
2015-06-10 08:01:44 <rnalrd> let me know if it works, so I can backport it to 3.2-stable
2015-06-10 08:03:24 <rnalrd> mosez, i'd not add all optional dependencies, but only the minimal required
2015-06-10 08:03:39 <rnalrd> since some depends are only needed if you enable certain modules
2015-06-10 08:06:33 <rnalrd> k, looks like perl-io is also needed
2015-06-10 08:14:09 <mosez> rnalrd: big thanks!
2015-06-10 08:18:08 <rnalrd> looks like that some modules are included in perl 5.22
2015-06-10 08:27:06 <ncopa> we should upgrade perl to 5.22
2015-06-10 08:27:32 <rnalrd> also 3.2-stable?
2015-06-10 08:27:59 <mosez> and you should upgrade pip as well :)
2015-06-10 08:28:11 <mosez> You are using pip version 6.1.1, however version 7.0.3 is available.
2015-06-10 08:28:11 <mosez> You should consider upgrading via the 'pip install --upgrade pip' command.
2015-06-10 08:35:23 <clandmeter> ncopa: done
2015-06-10 08:42:37 <mosez> what are these acf modules?
2015-06-10 08:43:59 <mosez> ah, alpine configuration framework
2015-06-10 08:44:42 <rnalrd> ncopa, I have perl-5.22.0 compiled locally
2015-06-10 08:44:47 <rnalrd> do we want me to push it?
2015-06-10 08:44:52 <rnalrd> do you*
2015-06-10 08:48:57 <ncopa> rnalrd: everything that has arch depedant perl modules needs to be rebuilt
2015-06-10 08:50:01 <rnalrd> hum, right
2015-06-10 08:51:53 <ncopa> that is grep '^arch="all"' perl-*/APK
2015-06-10 08:52:12 <ncopa> + some others, like perl-rrd, build from rrdtool
2015-06-10 08:52:36 <ncopa> grep perl-dev will probably give an indication
2015-06-10 08:52:51 <ncopa> rnalrd: do you want me do it?
2015-06-10 08:53:49 <ncopa> fwiw, i normally do this kind of things in a local git branch
2015-06-10 08:54:02 <rnalrd> do you commit in one go all packages?
2015-06-10 08:54:11 <ncopa> once i think i have all build tested i merge and rebase
2015-06-10 08:54:21 <ncopa> i think you syoud ideally push all in one go yes
2015-06-10 08:54:32 <ncopa> separate commits is ok
2015-06-10 08:54:43 <ncopa> but one push
2015-06-10 08:55:22 <ncopa> i suppose you could group the gommits too, like: main/per-*: rebuild against perl 5.22
2015-06-10 09:11:09 <rnalrd> i'll look at that later, got other stuff to do
2015-06-10 12:10:34 <StarWarsFan> hi@all
2015-06-10 12:10:52 <StarWarsFan> just a quick question: how about cross compiling on alpine?
2015-06-10 12:11:28 <StarWarsFan> is it possible to build i. e. for raspery a.s.o. on "real" hardware?
2015-06-10 12:13:25 <clandmeter> wandboard not real enough?
2015-06-10 12:14:21 <StarWarsFan> don't want to buy such board to see that it's not enough
2015-06-10 12:14:31 <StarWarsFan> and additionally it's another hardware to fail
2015-06-10 12:14:44 <StarWarsFan> if i can use existing hardware,
2015-06-10 12:14:50 <StarWarsFan> it would be much better
2015-06-10 12:15:21 <StarWarsFan> additonally i could build right on my own machine
2015-06-10 14:03:38 <clandmeter> ncopa: i think ^ should fix that font issue you were having
2015-06-10 14:03:53 <clandmeter> will check another time for fontconfig changes.
2015-06-11 07:48:10 <fabled> ncopa, new 3.18 is imminent. let's also bump the minimum x86 to 586 during that upgrade?
2015-06-11 07:49:51 <ncopa> in edge yes
2015-06-11 07:50:43 <ncopa> hum
2015-06-11 07:51:04 <ncopa> shared-mime-info is dog slow when there is lot of other io activity on server
2015-06-11 07:51:21 <fabled> still/again?
2015-06-11 07:51:28 <ncopa> looks like it takes ~5mins if kernel is compiled at same time
2015-06-11 07:51:33 <ncopa> still
2015-06-11 07:51:51 <ncopa> no, 2mins
2015-06-11 11:04:12 <ncopa> ScrumpyJack: could you please check that I applied all your sent patches?
2015-06-11 11:35:47 <xen_roger> hi guys,  got this error, but do not know what exactly goes wrong. "hostname:~# apk upgrade  1 errors; 13 MiB in 27 packages"
2015-06-11 11:35:52 <xen_roger> anyway to get what the error is?
2015-06-11 11:39:23 <ncopa> hum
2015-06-11 11:39:44 <ncopa> ttilley: llvm does not build with gcc5
2015-06-11 11:40:01 <ncopa> well, it seems to build, the make install does not appear to work
2015-06-11 11:40:28 <xen_roger> related package to my problem seems to be freeradius
2015-06-11 11:40:42 <xen_roger> can't be removed updated or installed
2015-06-11 11:41:06 <ncopa> xen_roger: apk version -l '?'
2015-06-11 11:41:33 <ncopa> does it show anything?
2015-06-11 11:41:52 <xen_roger> ncopa: "apk: option requires an argument: l"
2015-06-11 11:42:13 <xen_roger> sorry,
2015-06-11 11:42:15 <xen_roger> freeradius3-lib-3.0.3-r6                ?
2015-06-11 11:42:47 <ncopa> i know what the problem is
2015-06-11 11:42:59 <ncopa> freeradius3 got renamed to freeradius
2015-06-11 11:43:18 <xen_roger> ok, anyway to fix it?
2015-06-11 11:43:42 <ncopa> i suppose you could apk del freeradius3 && apk add freeraidus
2015-06-11 11:43:51 <ncopa> or in the opposite roder
2015-06-11 11:44:09 <ncopa> apk add -U -u freeradius
2015-06-11 11:44:19 <ncopa> apk del freeradius3
2015-06-11 11:44:25 <ncopa> bit it is a bug actually
2015-06-11 11:45:25 <xen_roger> ncopa: thank you! works
2015-06-11 11:45:31 <xen_roger> no upgrade error anymore
2015-06-11 12:00:46 <ncopa> i am fixing the root cause
2015-06-11 13:36:03 <ahills> It's exciting to live on the edge
2015-06-11 13:39:07 <clandmeter> untill we break it :)
2015-06-11 13:39:33 <ahills> That's what makes it exciting!
2015-06-11 13:40:41 <ahills> Sorry to anyone who was waiting for anything from me, I will be traveling through midsummer and won't be using a computer at all
2015-06-11 13:41:30 <clandmeter> ok let me hide a rm -rf / in a post-install somewhere :)
2015-06-11 13:42:38 <vkrishn> clandmeter, if its ram install... go ahead
2015-06-11 13:44:29 <vkrishn> no huf, and puf would blow the ram[s] stuffs off
2015-06-11 13:45:02 <ahills> clandmeter is the reason all my disks are mounted ro by default
2015-06-11 13:47:59 <vkrishn> ahills, if you do `lbu ci` it would remount , save and remount(ro) again
2015-06-11 13:48:58 <ahills> vkrishn: I mean even on disk installations
2015-06-11 13:50:44 <vkrishn> don't recall, have not done a full disk install for awhile
2015-06-11 13:56:03 <clandmeter> ncopa: we do not have xinitrc.d support?
2015-06-11 13:56:58 <ahills> Isn't that as simple as for dir in xinitrc.d/*; do . $dir; done?
2015-06-11 13:58:28 <clandmeter> ahills: it is
2015-06-11 13:59:09 <clandmeter> but im wondering if we prefered way to execute a script after X has started.
2015-06-11 13:59:51 <clandmeter> i can add a script to xinitrc.d, but if its not added to xinitrc, it wont work for everyone.
2015-06-11 14:00:40 <clandmeter> so i should probably ask it differently:
2015-06-11 14:00:57 <clandmeter> ncopa: we dont have xinitrc.d support, do we have something alterntive?
2015-06-11 14:04:17 <ncopa> clandmeter: i dont know. you'll have to read xinit ;)
2015-06-11 14:17:47 <ahills> Heh... Was just reading /etc/X11/xinit/xinitrc, why is it called xinitrc.cpp in the CVS header?
2015-06-11 14:18:25 <ahills> Most distributions have a modified version, you could probably steal it from Debian/Arch
2015-06-11 14:20:18 <clandmeter> yes, im reading what the best way is
2015-06-11 14:20:26 <clandmeter> some dont have it like arch
2015-06-11 14:20:40 <clandmeter> atleast i can only find refs for users xinitrc
2015-06-11 14:22:05 <ahills> Would you like a patch?
2015-06-11 14:38:10 <clandmeter> ahills: thx but i think this should do.
2015-06-11 14:40:31 <ahills> clandmeter: excellent, I can use the distro's xinitrc now
2015-06-11 14:40:43 <ahills> Why ?* instead of *?
2015-06-11 14:41:06 <ahills> Oops, that last ? is because it's a question
2015-06-11 14:47:19 <clandmeter> i guess so its minimal one char, i copied it from arch wiki.
2015-06-11 14:49:01 <ahills> I can't imagine the corner case... Maybe shells that don't glob properly?
2015-06-11 14:52:23 <clandmeter> i could have applied similar like fedora http://pkgs.fedoraproject.org/cgit/xorg-x11-xinit.git/tree/xinitrc-common#n53
2015-06-11 14:52:53 <clandmeter> but this one looked better, incase i overlooked something.
2015-06-11 15:03:22 <clandmeter> really... i need to go home.
2015-06-12 05:21:05 <fabled> hmmh
2015-06-12 05:21:18 <fabled> the new openssl releases have an ABI incompatibility
2015-06-12 05:22:08 <fabled> http://marc.info/?l=openssl-dev&m=143407129721271&w=2
2015-06-12 05:27:36 <kaniini> oh great
2015-06-12 05:27:52 <kaniini> so we get to look at openssl vs libressl again
2015-06-12 06:16:18 <ncopa> maybe we disable ssl3 compiletime
2015-06-12 06:30:23 <fabled> ncopa, ? we did that earlier
2015-06-12 06:30:41 <fabled> or did we
2015-06-12 06:45:36 <ncopa> i think we didnt
2015-06-12 06:47:27 <ncopa> disabling doing so will not break ABI will it?
2015-06-12 06:47:34 <ncopa> i think just do it now
2015-06-12 06:48:53 <ncopa> fabled: http://sprunge.us/Ejbh
2015-06-12 06:48:54 <ncopa> ok?
2015-06-12 06:49:12 <fabled> hmm
2015-06-12 06:49:19 <fabled> i wonder if it caused abi breakage
2015-06-12 06:49:21 <ncopa> apk still works...
2015-06-12 06:50:03 <fabled> checkapk?
2015-06-12 06:50:34 <ncopa> nothing.
2015-06-12 06:50:52 <ncopa> checkapk only check if filelist changes
2015-06-12 06:51:38 <ncopa> i can run it a bit on my devel box to see if anything broke
2015-06-12 06:53:17 <ncopa> i wonder if libressl is ABI compat
2015-06-12 07:19:56 <ncopa> sigh
2015-06-12 07:20:38 <ncopa> commit message and tag does not correspond
2015-06-12 07:20:49 <ncopa> i need coffe
2015-06-12 07:41:20 <clandmeter> fabled: on tmpfs install 3.2, when running apk i will get a: mount: mounting /dev/usbdisk on /media/usb failed: Resource busy
2015-06-12 07:41:30 <clandmeter> after which usb will be mount rw
2015-06-12 07:41:38 <fabled> modloop
2015-06-12 07:41:42 <fabled> clandmeter, i think it's kernel issue
2015-06-12 07:41:45 <fabled> when modloop is mounted
2015-06-12 07:41:56 <fabled> the kernel somehow makes the loop0 device be read-write
2015-06-12 07:42:17 <fabled> it has happened always
2015-06-12 07:42:36 <clandmeter> i cannot remember this happend in prev releases
2015-06-12 07:43:00 <clandmeter> shouldnt apk mount it rw, and afterwards ro again?
2015-06-12 07:43:06 <fabled> yes
2015-06-12 07:43:09 <fabled> the problem
2015-06-12 07:43:12 <fabled> apk mounts rw
2015-06-12 07:43:22 <fabled> *operation takes time, kernel remounts loop0 as rw*
2015-06-12 07:43:29 <fabled> apk finishes and remounts ro
2015-06-12 07:43:34 <fabled> but that fails since loop0 is now rw
2015-06-12 07:43:59 <fabled> actually that might bb mount bug
2015-06-12 07:44:12 <clandmeter> you are talking about loop0, thats modloop right?
2015-06-12 07:44:16 <fabled> yes
2015-06-12 07:44:35 <clandmeter> i dont get the reference to /media/usb?
2015-06-12 07:44:45 <clandmeter> i can still mount that ro if i want
2015-06-12 07:54:21 <fabled> hmm
2015-06-12 07:54:25 <fabled> weird
2015-06-12 08:08:39 <ncopa> strace?
2015-06-12 08:33:08 <fabled> i wonder what to do with the openssl mess
2015-06-12 08:33:18 <fabled> i'll probably upgrade by revert commit 2674af2f790aa1c869dcf5f2024e9eb6758edf56 that added the ABI incompat
2015-06-12 08:44:28 <clandmeter> fabled: http://sprunge.us/YhON
2015-06-12 08:45:03 <clandmeter> strace of apk update with resource busy
2015-06-12 08:48:02 <fabled> clandmeter, can you strace with -f; the actual error comes from "mount" that is fork+execve'd
2015-06-12 08:58:56 <clandmeter> fabled: http://sprunge.us/FYcI
2015-06-12 09:00:44 <ncopa> openssl annoys me
2015-06-12 09:00:55 <ncopa> maybe we should move to libressl?
2015-06-12 09:03:09 <clandmeter> seems kaniini wasnt convinced it is/was the right way early today in #alpine-linux
2015-06-12 09:04:30 <fabled> ncopa, you want me to add no-sslv3 too?
2015-06-12 09:06:13 <fabled> ok, openssl list woke up to the abi issue. let's wait a bit to see what they apply
2015-06-12 09:07:18 <fabled> clandmeter, what apk version is that?
2015-06-12 09:07:57 <clandmeter> apk-tools 2.6.1, compiled for x86_64.
2015-06-12 09:09:17 <fabled> ok
2015-06-12 09:09:21 <fabled> i think it's fixed in apk-tools.git
2015-06-12 09:09:30 <fabled> i should probably tag a new release and apply it to edge
2015-06-12 09:11:52 <fabled> clandmeter, can you try if 2.6.2 fixes it
2015-06-12 09:11:58 <fabled> i think it should
2015-06-12 09:12:05 <clandmeter> can i try the static one?
2015-06-12 09:12:10 <fabled> yes
2015-06-12 09:12:13 <clandmeter> ok
2015-06-12 09:14:14 <clandmeter> fabled: looks right
2015-06-12 09:24:50 <fabled> ncopa, look at http://sprunge.us/KhVH
2015-06-12 09:24:55 <fabled> it's the openssl upgrade commit
2015-06-12 09:25:01 <fabled> should we add no-sslv3 while at it?
2015-06-12 09:26:18 <ncopa> fabled: i'd like to that in separate commit
2015-06-12 09:26:30 <ncopa> so we can cherry-pick to 3.2-stable
2015-06-12 09:27:26 <fabled> ok
2015-06-12 09:27:29 <fabled> i go now
2015-06-12 09:27:37 <fabled> please push my commit if it's ok
2015-06-12 09:27:45 <fabled> though
2015-06-12 09:27:48 <fabled> it might need some testing
2015-06-12 09:27:53 <fabled> i did not really test the hmac patch
2015-06-12 09:28:11 <fabled> or should we wait until they fix the abi issue upstream?
2015-06-12 09:28:12 <ncopa> i'd like wait what openssl upstream says about it
2015-06-12 09:28:21 <fabled> yeah
2015-06-12 09:28:23 <ncopa> hum
2015-06-12 09:28:26 <ncopa> or actually
2015-06-12 09:28:29 <fabled> 1.0.1 is affected with the abi issue too
2015-06-12 09:28:32 <fabled> 1.0.0 is probably not
2015-06-12 09:28:38 <ncopa> we could probably use your fix
2015-06-12 09:28:44 <ncopa> and later replace with what upstream does
2015-06-12 09:29:09 <fabled> i think we use 1.0.1 or newer in all stable branches
2015-06-12 09:29:13 <ncopa> no-ssl3 can be added in separate commit
2015-06-12 09:29:32 <fabled> ok
2015-06-12 09:29:37 <fabled> i'll be back in 30mins or so
2015-06-12 09:29:57 <ncopa> i'll test your patch meanwhile
2015-06-12 11:31:35 <alacerda_> thijsschnitger, were you the one talking via twitter ?
2015-06-12 11:31:51 <thijsschnitger> that was my colleague
2015-06-12 11:32:11 <thijsschnitger> i'm looking into getting java8 running on alpine
2015-06-12 11:32:52 <thijsschnitger> i've read up on de libc issues but i was wondereing if i could somehow get openjdk running
2015-06-12 11:33:24 <thijsschnitger> although i don't know where to start, i really could use some pointers
2015-06-12 11:34:29 <clandmeter> thijsschnitger: http://git.alpinelinux.org/cgit/aports/tree/main/openjdk7
2015-06-12 11:37:11 <thijsschnitger> thanks, I've noticed support for 7 but I was wondering about 8.
2015-06-12 11:38:31 <clandmeter> you said "some pointers" :)
2015-06-12 11:39:31 <thijsschnitger> :D
2015-06-12 11:40:16 <clandmeter> fabled is the one who maintains it. you better ask when he is around.
2015-06-12 11:41:14 <Jamie_> helllo
2015-06-12 11:41:22 <Jamie_> I can see my colleague Thijs is in here now.
2015-06-12 11:41:46 <Jamie_> Basically, we are going to look into Alpine and write a bit about it.
2015-06-12 11:42:21 <alacerda> hi Jamie_, thijsschnitger, was asking about java8
2015-06-12 11:44:15 <thijsschnitger> ok clandmeter tnx
2015-06-12 13:06:13 <ScrumpyJack> i really ought to get twitter one daty
2015-06-12 13:06:17 <ScrumpyJack> daty?
2015-06-12 13:06:52 <ahills> Don't bother, it's a complete waste of time
2015-06-12 13:10:14 <ScrumpyJack> ahills: But people are talking about Alpine Linux on twitter, and only cool people talk about Alpine Linux, so twitter *must* be cool!
2015-06-12 13:14:05 <alacerda> ScrumpyJack, ahills We use twitter mainly to spread alpine's news like: new releases and some new packages.
2015-06-12 13:17:43 <ScrumpyJack> alacerda: we get it, we're not *that* retarded ;)
2015-06-12 13:18:44 <ScrumpyJack> hey, i though depends_dev was for packages to include if I was making a package-dev subpackage
2015-06-12 13:19:31 <ScrumpyJack> fcolista: thanks for your email - can you confirm?
2015-06-12 13:19:33 <clandmeter> yes, depends for pkgname-dev
2015-06-12 13:20:58 <ScrumpyJack> fcolista just told me off for putting -dev dependancies for a non-dev build into makedepends :)
2015-06-12 13:21:13 <ScrumpyJack> if that makes sense
2015-06-12 13:21:24 <fcolista> ScrumpyJack, yeah. Just didn't realize that you had not -dev package.
2015-06-12 13:22:01 <ScrumpyJack> fcolista: ok cool. thanks for pushing the patch
2015-06-12 13:22:02 <fcolista> makedepends has often $depends_dev in dependencies
2015-06-12 13:22:19 <fcolista> np
2015-06-12 13:22:25 <clandmeter> i guess there are lots of packages that have dep_dev deps but actually doesnt have a -dev pkg
2015-06-12 13:22:33 <fcolista> clandmeter, yes.
2015-06-12 13:22:55 <clandmeter> there are some recent changes to build
2015-06-12 13:32:17 <ScrumpyJack> ok here is another one
2015-06-12 13:33:02 <ScrumpyJack> abuild has pulled in like 50 -dev packages as dependancies for gthumb
2015-06-12 13:33:07 <ScrumpyJack> that can't be right
2015-06-12 13:34:07 <barthalion> sounds normal
2015-06-12 13:34:12 <barthalion> it's gnome project
2015-06-12 13:34:15 <ahills> It has a g in it
2015-06-12 13:34:32 <ahills> G is for gbloat
2015-06-12 13:34:43 <ahills> (Silent g)
2015-06-12 13:34:47 <barthalion> also it's one from thie ancient stuff developed since ever
2015-06-12 13:35:36 <clandmeter> ScrumpyJack: when building or when install gthumb?
2015-06-12 13:36:31 <ScrumpyJack> I'm testing the install of gthumb that i just abuilt
2015-06-12 13:36:53 <clandmeter> you added a -dev to depends probably?
2015-06-12 13:37:06 <ahills> apk info -R lists direct dependencies
2015-06-12 13:37:11 <clandmeter> it should be added to makedeps
2015-06-12 13:37:53 <ScrumpyJack> depends is empty
2015-06-12 13:38:13 <ScrumpyJack> http://sprunge.us/RdBW
2015-06-12 13:39:03 <clandmeter> you dont need to add \ in makedepends
2015-06-12 13:40:20 <ScrumpyJack> ok. doesnt hurt. i'll remove them. but why have i pulled in a zillion -dev deps?
2015-06-12 13:40:36 <clandmeter> it must be one of the deps that pulls them in
2015-06-12 13:40:45 <clandmeter> so its broken somwheere else
2015-06-12 13:41:46 <ScrumpyJack> i'll tinker
2015-06-12 13:42:19 <ScrumpyJack> how do i get rid of the general dependancies of certain gnome apps on org.freedesktop.DBus ?
2015-06-12 13:42:49 <ScrumpyJack> this sort of thing: Gtk-WARNING **: Failed to register client: GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown
2015-06-12 13:47:55 <alacerda> kkk ScrumpyJack sorry wasn't my intention to say *that* :)
2015-06-12 13:54:25 <fabled> ok
2015-06-12 13:54:57 <fabled> seems openssl committed a fix
2015-06-12 13:54:59 <fabled> for the abi issue
2015-06-12 13:55:05 <fabled> i'll take that and push new openssl to edge
2015-06-12 13:55:31 <ahills> Can't wait to see what new vulnerabilities that simple fix brings
2015-06-12 13:55:46 <ahills> OpenSSL is almost as exciting as Microsoft
2015-06-12 13:57:23 <ScrumpyJack> fabled: your patch?
2015-06-12 13:57:46 <fabled> ScrumpyJack, slightly improved version of it: https://github.com/openssl/openssl/commit/1030f89f5ea238820645e3d34049eb1bd30e81c4.patch
2015-06-12 13:58:31 <clandmeter> ncopa: please revert that tvheadend commit
2015-06-12 13:58:55 <fabled> clandmeter, why?
2015-06-12 13:59:06 <clandmeter> 4.1 is not stable
2015-06-12 13:59:11 <fabled> ah
2015-06-12 13:59:14 <clandmeter> its writen in the apkbuild
2015-06-12 13:59:39 <clandmeter> its the second time its pushed to 4.1
2015-06-12 13:59:47 <clandmeter> or somebody must have a godo reason for it.
2015-06-12 14:01:43 <ncopa> sorry
2015-06-12 14:01:51 <ncopa> clandmeter: do you want take over maintenance?
2015-06-12 14:01:56 <ncopa> for tvheadend?
2015-06-12 14:01:59 <clandmeter> sure
2015-06-12 14:02:04 <clandmeter> i push it regular
2015-06-12 14:02:19 <clandmeter> when needed ofc :)
2015-06-12 14:03:10 <ncopa> thanks!
2015-06-12 14:05:06 <fabled> ncopa, i have 1.0.0n upgrade commit for 3.1, 3.0 and 2.7 ready too
2015-06-12 14:05:18 <fabled> i'll push 3.1?
2015-06-12 14:05:31 <fabled> we can then cherry pick edge -> 3.0; and 3.1 -> 3.0/2.7
2015-06-12 14:07:40 <ncopa> ok
2015-06-12 14:07:59 <ncopa> what about v3.2?
2015-06-12 14:09:19 <fabled> err
2015-06-12 14:09:23 <fabled> ment edge -> 3.2
2015-06-12 14:09:30 <fabled> just cherry-picks
2015-06-12 14:10:04 <ScrumpyJack> gthumb doesn't run :( hey-ho
2015-06-12 14:12:36 <fabled> ncopa, can you cherry pick openssl from edge -> 3.2; and from 3.1 -> 3.0 and 2.7 ?
2015-06-12 14:12:39 <fabled> i need to go
2015-06-12 14:15:52 <ncopa> i'll do
2015-06-12 14:15:54 <ncopa> thanks!
2015-06-12 14:28:39 <ScrumpyJack> do we have sqlite3 libs?
2015-06-12 14:30:44 <AmatCoder> ScrumpyJack: yes
2015-06-12 14:31:17 <AmatCoder> sqlite-libs is 3.x version
2015-06-12 14:35:02 <ScrumpyJack> thank you. this pkg-config can't find anything at abuild time
2015-06-12 14:54:51 <ncopa> abuild?
2015-06-12 15:08:13 <ScrumpyJack> is the a webkitgtk3?
2015-06-12 15:10:13 <ScrumpyJack> hmm
2015-06-12 15:11:52 <ScrumpyJack> webkitgtk is for gtk3
2015-06-12 15:14:17 <ScrumpyJack> so, what's the difference between gtksourceview2-2.10.5 and gtksourceview-2.10.4
2015-06-12 15:18:51 <ScrumpyJack> ncopa: shouldn't either gtksourceview or gtksourceview2 pull in gtk+3.0 ?
2015-06-12 15:19:43 <ScrumpyJack> they both pull in gtk+2.0
2015-06-12 15:21:41 <ScrumpyJack> (nope, neither gtksourceview or gtksourceview2 builds with gtk+3.0
2015-06-12 15:24:12 <ScrumpyJack> just ignore my rambling (you already are) :)
2015-06-12 15:24:28 <AmatCoder> ScrumpyJack: to build with gtk+3.0 you would need gtksourceview 3.x.x version
2015-06-12 15:28:30 <AmatCoder> 3.16.1 is the last
2015-06-12 15:29:33 <AmatCoder> but yes, gtksourceview or gtksourceview2 packages are equal...
2015-06-12 15:32:28 <ScrumpyJack> it's building as we type
2015-06-12 15:50:04 <ScrumpyJack> where can i get xmllint from?
2015-06-12 15:50:34 <ScrumpyJack> i was hoping libxml2
2015-06-12 15:53:37 <AmatCoder> libxml2-utils
2015-06-12 15:55:43 <ScrumpyJack> ay, nice one
2015-06-12 15:56:02 <AmatCoder> :)
2015-06-12 16:21:40 <kaniini> clandmeter: my view is that with libressl we are just trading in one pile of crap for another pile crap
2015-06-12 16:22:09 <kaniini> clandmeter: and that ultimately using libressl may cause people to be complacent and not do "the right thing" (get rid of use of those APIs altogether)
2015-06-12 16:22:58 <kaniini> clandmeter: it may be that libressl is a good bandaid, but we really just need to push for an end to that entire ecosystem
2015-06-15 07:10:02 <clandmeter> morning
2015-06-15 07:10:49 <clandmeter> ncopa: on my laptop, wpa is started after networking.
2015-06-15 07:20:41 <clandmeter> lm_sensors also seem broken
2015-06-15 07:49:46 <ncopa> clandmeter: how is lm_sensors broken?
2015-06-15 07:50:06 <ncopa> i think you can do: rc_before="networking" in /etc/conf.d/wpa_supplicant
2015-06-15 07:50:27 <ncopa> but it might be an idea to always start wpa_supplicant before networking
2015-06-15 07:50:39 <ncopa> clandmeter: thanks for the font packages btw
2015-06-15 08:02:25 <clandmeter> ncopa: it doesnt generate a confd file
2015-06-15 08:02:32 <clandmeter> so it doesnt want to start
2015-06-15 08:03:10 <clandmeter> ncopa: regarding fonts, let me know if it works as expected for you.
2015-06-15 08:03:59 <clandmeter> there should be some replacement fonts now so size should be the same as their non-free versions.
2015-06-15 08:09:01 <clandmeter> ncopa: regarding sensors, my notbook, it only finds coretemp. maybe its not enough to generate the confd file.
2015-06-15 08:17:43 <vkris> bump php-phalcon v2.0.3 pls
2015-06-15 08:21:07 <vkris> probably it does not have other sensors
2015-06-15 08:24:29 <clandmeter> vkris: how is irrsi?
2015-06-15 08:24:42 <vkris> cool !
2015-06-15 08:25:09 <vkris> but I miss channel-name tabs
2015-06-15 08:25:34 <clandmeter> there are some hacks to add it i think
2015-06-15 08:25:49 <clandmeter> but i never minded that much
2015-06-15 08:26:51 <vkris> cannot figure out if a msg popped on non active channel
2015-06-15 08:35:12 <clandmeter> vkris: what do you mean?
2015-06-15 08:35:23 <clandmeter> you dont see the numbers in Act: ?
2015-06-15 08:35:47 <ScrumpyJack> clandmeter: Alpine Linux on your laptop? neat!
2015-06-15 08:36:12 <clandmeter> and hot :)
2015-06-15 08:36:33 <clandmeter> wonder if i can get it to cool down a bit
2015-06-15 08:36:39 <ScrumpyJack> which biuld? setup-box or alpine-desktop or hand rolled?
2015-06-15 08:36:40 <clandmeter> on winddows, its much cooler
2015-06-15 08:37:20 <clandmeter> i think setup-desktop. but it needs more work to get it working
2015-06-15 08:37:31 <clandmeter> running it from an usb3 stick :)
2015-06-15 08:37:50 <ScrumpyJack> is suspend working?
2015-06-15 08:38:01 <clandmeter> it should, but didnt try
2015-06-15 08:38:07 <clandmeter> it works on my desktop
2015-06-15 08:38:39 <ScrumpyJack> i mean suspend when you close the lid
2015-06-15 08:38:47 <clandmeter> i know what you mean
2015-06-15 08:38:52 <ScrumpyJack> :)
2015-06-15 08:38:52 <clandmeter> but i think you need to set that up
2015-06-15 08:39:00 <clandmeter> acpi events right?
2015-06-15 08:39:07 <ScrumpyJack> something like that
2015-06-15 08:39:40 <ScrumpyJack> there is something else, as configuring acpi wasn't enough
2015-06-15 08:39:51 <clandmeter> you need pm-utils
2015-06-15 08:39:53 <clandmeter> probably
2015-06-15 08:41:33 <ScrumpyJack> there was some talk about it earlier, but i wasn't listening :)
2015-06-15 08:42:51 <vkris> hmm... Act:... just wondered how to interpret it ;)
2015-06-15 08:44:21 <clandmeter> vkris: RTFM ;-)
2015-06-15 08:44:29 <vkris> onto it
2015-06-15 08:44:45 <clandmeter> 3 states
2015-06-15 08:44:58 <clandmeter> lightblue is part/join
2015-06-15 08:45:02 <clandmeter> white is msg
2015-06-15 08:45:09 <clandmeter> red is mention
2015-06-15 08:46:13 <barthalion> cool kids use weechat nowadays
2015-06-15 08:46:38 <barthalion> the bee's knees
2015-06-15 08:47:27 <clandmeter> i thought it didnt work on alpine?
2015-06-15 08:47:32 <clandmeter> something with terms
2015-06-15 08:49:04 <barthalion> no idea, I don't use it on alpine
2015-06-15 08:53:56 <ahills> I use weechat on Alpine
2015-06-15 08:56:31 <ahills> clandmeter: in about a week I can share some of my setup with you, all that's missing currently is suspend to disk
2015-06-15 08:56:52 <ahills> The wiki article for lid suspend is still relevant
2015-06-15 08:57:43 <vkris> http://quadpoint.org/articles/irssisplit/ <- statusbar show channel names
2015-06-15 09:04:50 <ScrumpyJack> weechat pulls in perl and python, and irssi doesn't, which is enough for me to stick with irssi for now :)
2015-06-15 09:09:07 <barthalion> irssi probably pulls perl
2015-06-15 09:12:11 <ScrumpyJack> not in edge. i don't know about stable
2015-06-15 10:29:21 <clandmeter> barthalion: irssi dont pull perl, i think its a subpkg
2015-06-15 12:22:45 <ncopa> i switched back to hexchat...
2015-06-15 13:51:13 <vkris> in http://pkgs.alpinelinux.org/contents?pkgname=irssi-perl&arch=x86 /usr/share/irssi/scripts/msg-event.pl seems to be scripts/example directory in source
2015-06-15 15:21:33 <vkris> irssi pkg does not have /usr/bin/botti ?
2015-06-15 20:18:25 <pnutzh4x0r> clandmeter: another workaround re: wpa_supplicant is to add it to the boot runlevel rather than the default
2015-06-15 20:19:07 <pnutzh4x0r> i ran into the same problem, and doing rc-update add wpa_supplicant boot allowed wpa_supplicant to start before networking
2015-06-15 20:19:27 <pnutzh4x0r> i wonder if it is because networking is also in boot
2015-06-15 20:19:57 <pnutzh4x0r> and thus both have to be in the same runlevel for the dependencies to work
2015-06-15 20:24:34 <clandmeter> pnutzh4x0r: i was thinking the same but didnt try it yet. will do when im behind my nb again.
2015-06-15 21:38:03 <BitL0G1c> clandmeter / pnutzh4x0r - for wlan putting the settings in /etc/network/interfaces works every time http://linuxcommando.blogspot.nl/2013/10/how-to-connect-to-wpawpa2-wifi-network.html & use the settings mentioned further down by "Emerson Prado"
2015-06-16 00:42:53 <pnutzh4x0r> BitL0G1c: thnx, i ll look into it
2015-06-16 00:47:20 <BitL0G1c> shortened instructions http://it-offshore.co.uk/linux/alpine-linux/46-alpine-linux-automatic-wireless-configuration
2015-06-16 07:09:19 <fabled> ncopa, qt5 packaging now pulls inqt5-qtbase-dev ?
2015-06-16 07:10:39 <fabled> ERROR: qt5-qtbase-dev-5.4.2-r0: trying to overwrite usr/include/QtTest/QSignalSpy owned by qt-dev-4.8.6-r2.
2015-06-16 07:10:56 <ncopa> hum
2015-06-16 07:11:02 <fabled> qt5-qtbase-dev-5.4.2-r0 is required by:
2015-06-16 07:11:02 <fabled> qt5-qtbase-5.4.2-r0
2015-06-16 07:11:40 <ncopa> depend = qt5-qtbase-dev=5.4.2-r0
2015-06-16 07:11:43 <ncopa> seems so yes
2015-06-16 07:11:46 <ncopa> i wonder why
2015-06-16 07:14:03 <ncopa> aha
2015-06-16 07:14:10 <ncopa> there are some symlink
2015-06-16 07:14:15 <ncopa> that point to something in -dev
2015-06-16 07:16:08 <ncopa> found it
2015-06-16 07:16:40 <ncopa> all the usr/bin/*-qt5 symlinks points to binaries in usr/lib/qt5/bin/ in -dev package
2015-06-16 07:19:08 <ncopa> im on it
2015-06-16 07:53:42 <ScrumpyJack> i'm going to update live 3.1 to 3.2. can i avoid upgrading the kernel?
2015-06-16 07:57:10 <ncopa> yes. just changing v3.1 to v3.2 in /etc/apk/repositories should do that
2015-06-16 08:02:19 <ScrumpyJack> sure, but that won't upgrade the kernel?
2015-06-16 08:20:39 <ncopa> not if its a tmpfs install
2015-06-16 08:53:42 <ncopa> i finally fixed http://git.alpinelinux.org favicon
2015-06-16 08:53:46 <ncopa> i also changed the title
2015-06-16 08:54:05 <ncopa> so its easier to find the correct tab
2015-06-16 09:35:42 <fcolista> ScrumpyJack, it's fine if i add you as maintainer of corebird package?
2015-06-16 10:04:55 <barthalion> no cookie for fcolista
2015-06-16 10:56:45 <fcolista> boom
2015-06-16 10:56:47 <fcolista> what's up
2015-06-16 10:56:48 <fcolista> ?
2015-06-16 10:57:12 <fcolista> problem with sphinx
2015-06-16 10:58:39 <fcolista> parallel build issue again?
2015-06-16 11:18:00 <vkris> https://github.com/rofl0r/jobflow seems nice
2015-06-16 11:22:28 <barthalion> no commits since a year
2015-06-16 12:05:58 <vkris> ScrumpyJack: have you tried snownews ?
2015-06-16 12:07:45 <vkris> earlier I could not get this "http://www.php.net/feed.atom" parsed
2015-06-16 12:22:16 <vkris> bump weechat to v1.2 pls
2015-06-16 12:22:53 <vkris> oh, it already
2015-06-16 12:26:58 <vkris> proxychains-ng to v4.10 , wow we missed this for 2cycles
2015-06-16 12:35:20 <ncopa> i am looking at gitolite
2015-06-16 12:35:33 <ncopa> i wonder if i should just migrate it all right away
2015-06-16 12:36:00 <ncopa> or maybe set up the user repos first
2015-06-16 12:36:33 <barthalion> I'd start from user repos (with wild repos enabled)
2015-06-16 12:36:49 <barthalion> are we going to give more people access to aports now?
2015-06-16 12:37:34 <ncopa> i want make it easy to give people access
2015-06-16 12:39:58 <ncopa> there are some user account that has not be used for years
2015-06-16 12:40:02 <ncopa> need to clean up that too
2015-06-16 12:40:36 <barthalion> you should also let people delete their wild repositories
2015-06-16 12:40:44 <ncopa> yes
2015-06-16 12:41:00 <barthalion> and enable possibility to set WRITERS for those
2015-06-16 12:41:18 <ncopa> how do i do that?
2015-06-16 12:41:48 <barthalion> http://gitolite.com/gitolite/user.html 4.1
2015-06-16 12:43:52 <ncopa> barthalion: do you want user/bpiotrowski/ or user/barthalion/
2015-06-16 12:44:07 <ncopa> cd ..
2015-06-16 12:44:34 <barthalion> I think I have bpiotrowski now?
2015-06-16 12:44:43 <ncopa> yes
2015-06-16 12:44:57 <barthalion> so it's good as is
2015-06-16 12:52:22 <ncopa> barthalion: i gave you access i think
2015-06-16 12:52:34 <ncopa> try: git clone git@git.alpinelinux.org:gitolite-admin
2015-06-16 12:54:28 <barthalion> yep, it works
2015-06-16 12:55:33 <ncopa> to allow people delete wild repos, i add "D = CREATOR"?
2015-06-16 12:57:02 <barthalion> hm, no
2015-06-16 12:57:06 <barthalion> C creator is enough
2015-06-16 12:57:16 <barthalion> you need to enable 'D' command in gitolite config though
2015-06-16 12:57:42 <clandmeter> ncopa: are you going to remove regular ssh access?
2015-06-16 12:57:49 <ncopa> i think not
2015-06-16 12:57:54 <ncopa> at least not initially
2015-06-16 12:57:56 <clandmeter> ok
2015-06-16 12:58:04 <clandmeter> leave me out of the removal please :)
2015-06-16 12:58:28 <ncopa> clandmeter: you'd be one of the last removed :)
2015-06-16 12:58:33 <clandmeter> i use it for distfiles
2015-06-16 12:58:36 <ncopa> yes
2015-06-16 12:58:53 <clandmeter> and public_html a bit
2015-06-16 12:58:54 <ncopa> i might do a split of dev.a.o and git.a.o
2015-06-16 12:59:03 <clandmeter> makes sence
2015-06-16 12:59:21 <ncopa> i might remove full ssh access to git.a.o, but let devs have ssh access to dev.a.o
2015-06-16 12:59:27 <ncopa> for the things you mentioned
2015-06-16 13:01:11 <ncopa> i might disable password logins though
2015-06-16 14:10:51 <ncopa> i have added user repos
2015-06-16 14:11:09 <ncopa> support for user repos
2015-06-16 14:11:30 <ncopa> clandmeter: can you try push something to git@git.alpinelinux.org:user/clandmeter/*
2015-06-16 14:11:44 <clandmeter> k
2015-06-16 14:17:00 <clandmeter> workded but i dont see it
2015-06-16 14:18:06 <clandmeter> ncopa: should i do something specific to find it on cgit?
2015-06-16 14:18:36 <ncopa> no i havent configured cgit yet
2015-06-16 14:19:17 <clandmeter> how can i remove my_l33t_repo.git?
2015-06-16 14:19:21 <clandmeter> :)
2015-06-16 14:20:27 <ncopa> http://gitolite.com/gitolite/gitolite.html#deleting-a-wild-repo
2015-06-16 14:20:48 <ncopa> hum
2015-06-16 14:20:55 <ncopa> i need enable the 'D' command
2015-06-16 14:22:27 <clandmeter> i dont get it?
2015-06-16 14:22:43 <clandmeter> ssh git@git@git.alpinelinux.org D -h should show a help or something?
2015-06-16 14:22:56 <ncopa> yes
2015-06-16 14:23:00 <ncopa> but i have to enable it
2015-06-16 14:23:07 <ncopa> trying to figure out how...
2015-06-16 14:23:42 <clandmeter> it asks for a password
2015-06-16 14:23:56 <ncopa> hum
2015-06-16 14:24:01 <ncopa> it should have used your ssh key
2015-06-16 14:24:17 <clandmeter> thats what i thought
2015-06-16 14:25:06 <clandmeter> oh now i get more
2015-06-16 14:26:08 <clandmeter> oh lol
2015-06-16 14:26:17 <clandmeter> i used double git@ :)
2015-06-16 14:26:49 <ncopa> ok, 'D' is enabled
2015-06-16 14:27:46 <clandmeter> 'user/clandmeter/my_l33t_repo' is locked!
2015-06-16 14:28:11 <clandmeter> ah got it
2015-06-16 14:28:17 <clandmeter> i can unlock myself
2015-06-16 14:28:28 <clandmeter> it tries to prevent cutting yourself
2015-06-16 14:30:51 <ncopa> 'user/ncopa/mytest' moved to trashcan
2015-06-16 14:30:59 <ncopa> finally managaed to delete it
2015-06-16 14:36:09 <clandmeter> seems to work ok
2015-06-16 14:47:03 <ncopa> the repos now also show up on git.alpinelinux.org/cgit
2015-06-16 14:49:02 <clandmeter> ncopa: previous ones are not mananged by gitolite?
2015-06-16 14:50:07 <ncopa> no
2015-06-16 14:50:27 <ncopa> the only managed by gitolite is currently user/*
2015-06-16 15:20:19 <barthalion> ncopa: if you could give me ssh access to git account, I can help with anything you need
2015-06-16 16:29:02 <ScrumpyJack> vkris: yes. there is a problem with the charset and ncurses. i'll rebuld it as some stage
2015-06-16 16:29:23 <ScrumpyJack> but it's better than newbeuter that doesn't even start :)
2015-06-16 16:29:37 <ScrumpyJack> newsbeuter even
2015-06-16 16:40:47 <barthalion> ncopa: http://comments.gmane.org/gmane.comp.embedded.openwrt.devel/32651
2015-06-16 17:56:46 <vkris> ScrumpyJack: http://i.imgur.com/tHeGCxI.png, except for unaligned columns, seems to work in AL
2015-06-16 17:58:07 <vkris> that was via ssh login to AL box
2015-06-16 18:55:55 <alacerda> We are on MATE's main page: http://mate-desktop.org/ "Which distributions support MATE?"
2015-06-16 18:56:04 <alacerda> :D
2015-06-16 18:56:06 <alacerda> \0/
2015-06-16 19:01:55 <alacerda> clandmeter, can you apply this please? http://sprunge.us/KLOe
2015-06-16 19:01:55 <alacerda>  
2015-06-16 19:06:26 <alacerda> or ncopa ^^
2015-06-16 21:02:43 <clandmeter> alacerda: mate-icon-theme*: Found /usr/share/locale
2015-06-16 22:43:02 <shafire> hi
2015-06-17 07:43:10 <ncopa> I am moving the user repos to gitolite now
2015-06-17 07:55:05 <ncopa> oh no...
2015-06-17 07:55:17 <ncopa> i think i killed a few repos
2015-06-17 07:55:20 <ncopa> user repos
2015-06-17 07:55:24 <ncopa>  /o\
2015-06-17 07:56:10 <barthalion> lol
2015-06-17 07:56:13 <barthalion> rest in peace
2015-06-17 07:56:23 <ncopa> well
2015-06-17 07:56:31 <ncopa> they were bad to begin with
2015-06-17 07:56:39 <ncopa> the ones that was not bare repos
2015-06-17 07:57:48 <ncopa> i did: git clone --bare --shared repo/ repo.git
2015-06-17 07:57:50 <ncopa> then moved that
2015-06-17 07:57:56 <ncopa> but now those are gone :-(
2015-06-17 07:58:10 <barthalion> you know what they say
2015-06-17 07:58:13 <barthalion> shit happens
2015-06-17 07:58:33 <barthalion> there are big chances that if these were actually used, someone has a copy
2015-06-17 07:58:53 <barthalion> and we can also set up mirroring one day to github or another gitolite instance
2015-06-17 08:08:44 <fabled> ncopa, i think issue is with that '--shared'
2015-06-17 08:08:58 <fabled> it makes git not copy all data
2015-06-17 08:09:11 <fabled> instead it uses the original tree as local data source
2015-06-17 08:09:23 <ncopa> aw
2015-06-17 08:09:24 <fabled> since you deleted the original. itbroke.
2015-06-17 08:09:34 <ncopa> ok. thanks for the explanation
2015-06-17 08:09:39 <ncopa> makes sense
2015-06-17 08:09:59 <fabled> from man:
2015-06-17 08:09:59 <fabled>    --shared, -s
2015-06-17 08:10:00 <fabled>            When the repository to clone is on the local machine, instead of
2015-06-17 08:10:00 <fabled>            using hard links, automatically setup .git/objects/info/alternates
2015-06-17 08:10:00 <fabled>            to share the objects with the source repository. The resulting
2015-06-17 08:10:00 <fabled>            repository starts out without any object of its own.
2015-06-17 08:10:18 <ncopa> yeah
2015-06-17 08:10:24 <ncopa> that was my mistake
2015-06-17 08:10:25 <fabled> it has also a big fat NOTE that it's possibly dangerous :)
2015-06-17 08:10:26 <ncopa> well
2015-06-17 08:10:45 <Orochimarufan> what happened? :O
2015-06-17 08:10:45 <ncopa> mistake was that i purged old
2015-06-17 08:11:02 <ncopa> 3 user repos was not 'bare' repos
2015-06-17 08:11:16 <ncopa> during the conversion to bare repo i managed to purge the repo
2015-06-17 08:11:28 <Orochimarufan> ouch
2015-06-17 08:12:44 <Orochimarufan> stupid things like that seem to happen way too often -.-
2015-06-17 08:18:12 <Orochimarufan> Is there any sane way to get nsswitch functionality with musl?
2015-06-17 08:26:46 <ncopa> :)
2015-06-17 08:27:06 <ncopa> Is there any sane way to get <something totally insane> functionality with musl?
2015-06-17 08:27:21 <ncopa> Orochimarufan: i think you can ask in #musl
2015-06-17 08:27:58 <ncopa> i doubt there are any sane ways to implement nsswitch
2015-06-17 08:28:06 <Orochimarufan> I don't think nsswitch is <something totally insane> but whatever
2015-06-17 08:28:30 <ncopa> well, the functionality is maybe not insane
2015-06-17 08:28:32 <Orochimarufan> I'm thinking it should be possible to LD_PRELOAD or something, right?
2015-06-17 08:28:49 <ncopa> but the dlopen way they do it is very risky
2015-06-17 08:29:15 <ncopa> i dont think you can do it with LD_PRELOAD with musl
2015-06-17 08:29:28 <Orochimarufan> I don't really care HOW they do it, I just need some way to get my stuff into getent
2015-06-17 08:29:48 <ncopa> understand
2015-06-17 08:29:52 <Orochimarufan> preferrably without modifying /etc/{passwd,group}
2015-06-17 08:30:03 <ncopa> ask in #musl
2015-06-17 08:30:23 <ncopa> and fwiw, i am a bit interested in similar functionality
2015-06-17 08:30:50 <ncopa> Orochimarufan: where do you have your db? in ldap?
2015-06-17 08:30:55 <Orochimarufan> winbind
2015-06-17 08:31:32 <Orochimarufan> Samba 4.2 seems to require libnss-winbind to work properly for smbd to work as member server
2015-06-17 08:32:47 <ncopa> there are no trivial way to do it afaik
2015-06-17 08:33:19 <Orochimarufan> right now, I'm having a cronjob update /etc/...
2015-06-17 08:33:31 <Orochimarufan> It seems less-than-ideal to me
2015-06-17 08:33:36 <ncopa> yeah
2015-06-17 08:50:26 <ncopa> cloning git://git.alpinelinux.org/aports is broken til i have moved all the repos
2015-06-17 11:04:53 <alacerda> thanks clandmeter ;)
2015-06-17 11:15:29 <barthalion> ncopa: wrt shell access: http://gitolite.com/gitolite/sts.html – 4.1
2015-06-17 11:17:25 <ncopa> hum
2015-06-17 11:18:28 <ncopa> so you can log in with ssh git@git.a.o?
2015-06-17 11:18:37 <barthalion> yeah
2015-06-17 11:19:04 <barthalion> also we will probably want to do some redis caching
2015-06-17 11:19:24 <ncopa> ok? is is that slow?
2015-06-17 11:19:26 <barthalion> (rather later than soon)
2015-06-17 11:19:45 <barthalion> it will probably get slower with number of repositories growing
2015-06-17 12:39:53 <ncopa> clandmeter: i might have broken patchwork...
2015-06-17 12:40:07 <ncopa>  crond: USER clandmeter pid 22454 cmd cd /home/clandmeter/cgit/patchwork.git && git fetch -q --all -p
2015-06-17 12:40:26 <ncopa> the /home/*/cgit dirs has be replaced with a file
2015-06-17 13:37:37 <clandmeter> i dont think so
2015-06-17 13:37:55 <clandmeter> atleast, i am pushing from redmine like yo u
2015-06-17 13:38:02 <clandmeter> did you break this?
2015-06-17 13:38:08 <clandmeter> ncopa: ^
2015-06-17 13:39:30 <ncopa> i didnt touch redmine
2015-06-17 13:39:35 <ncopa> i just noticed there is a cronjob
2015-06-17 13:39:44 <ncopa>  cd /home/clandmeter/cgit/patchwork.git && git fetch -q --all -p
2015-06-17 13:39:50 <ncopa> that will fail i think
2015-06-17 14:16:07 <ncopa> moving aports now
2015-06-17 14:16:16 <ncopa> hooks should be ok more or less
2015-06-17 14:25:05 <ncopa> ok
2015-06-17 14:25:34 <ncopa> git commits are not echoed in this channel anymore
2015-06-17 14:26:25 <ncopa> do we want git commits echoed in this channel?
2015-06-17 14:47:16 <ncopa> i will set up the auto generation of release tarballs
2015-06-17 15:02:00 <clandmeter> ncopa: ah now i got it
2015-06-17 15:02:06 <clandmeter> you can remove it
2015-06-17 15:02:12 <clandmeter> i dont need that repo
2015-06-17 15:39:51 <stateless> ncopa, http://dl.2f30.org/releases/fortify-headers-0.6.tar.gz
2015-06-17 16:52:45 <alacerda> clandmeter or ncopa, please: http://sprunge.us/TgAg
2015-06-17 16:52:56 <alacerda> thanks in advance
2015-06-17 19:54:23 <clandmeter> ncopa: maybe we should have #alpine-aports? which displays commits and buildstatus?
2015-06-17 19:58:04 <alacerda> clandmeter, +1
2015-06-17 20:02:00 <clandmeter> alacerda: done
2015-06-17 20:02:05 <clandmeter> although no msg here ;-)
2015-06-17 21:24:41 <mitchty> i have a question on a port of ghc to alpine that i have
2015-06-17 21:24:51 <mitchty> i'm up to the point that i have apk's
2015-06-17 21:25:00 <mitchty> and it works or seems to at least
2015-06-17 21:25:06 <mitchty> at least for a version 0
2015-06-17 21:25:38 <mitchty> but given to build it you need to use it to build itself, how would that work when doing a request to add it into the package repo even as testing?
2015-06-17 21:26:24 <mitchty> i have the whole thing setup in 2 dockerfiles, one to cross compile a ghc targetting musl libc, and another dockerfile to build up to the apk within alpine itself
2015-06-17 21:26:37 <mitchty> well more a bootstrap apk
2015-06-17 21:26:58 <mitchty> or would this be better asked on the email lists?
2015-06-17 22:29:23 <clandmeter> mitchty: i think this was disccused recently on irc
2015-06-17 22:31:04 <clandmeter> the idea was to have 2 apk's and documentation on the wiki about how to bootstrap when needed for other archs.
2015-06-17 22:31:09 <clandmeter> but i could be wrong
2015-06-17 22:31:48 <mitchty> ah, k
2015-06-17 22:31:49 <clandmeter> ml sounds like the best way to discuss this.
2015-06-17 22:31:57 <mitchty> yeah the wiki pages ports have way out of date info
2015-06-17 22:32:03 <clandmeter> i think we have around 3 ppl who can comment on this.
2015-06-17 22:32:05 <mitchty> using chroots and all sorts of nonsense
2015-06-17 22:32:33 <mitchty> so my strategy to get ghc building was to use the sabotage linux musl gcc cross compiller
2015-06-17 22:32:44 <mitchty> should be easy to get to other arches that way
2015-06-17 22:33:03 <mitchty> only did x86_64 to start mostly because thats the one i needed
2015-06-17 22:33:10 <mitchty> https://github.com/mitchty/alpine-linux-ghc-bootstrap
2015-06-17 22:33:22 <mitchty> everything is in there at the moment
2015-06-17 22:33:56 <mitchty> wouldn't be too hard to get to a better more documented state
2015-06-17 22:35:47 <mitchty> but i got 7.10.1 built right before 7.10.2 is to come out, i'll ask on the ml probably late tonight, have some prior commitments shortly thanks clandmeter
2015-06-17 22:36:10 <mitchty> should i do alpine devel? somewhat unclear which is the proper ml
2015-06-17 22:36:28 <clandmeter> yes
2015-06-17 22:37:08 <clandmeter> you can keep an eye out for fabled, ncopa and kaniini
2015-06-17 22:37:32 <clandmeter> only thing i boostrapped was something from twitter :)
2015-06-17 22:37:40 <mitchty> lol
2015-06-17 22:38:05 <mitchty> its been a fun 2 weeks of looking cross eyed at things
2015-06-17 22:38:37 <mitchty> gah, that apkbuild didn't get into my git commit
2015-06-17 22:39:59 <mitchty> ok well need to jet for my spiel, will send out an email later, by that time this build might be done :)
2015-06-18 00:22:34 <kaniini> hi
2015-06-18 00:24:11 <kaniini> mitchty: so some initial review
2015-06-18 00:24:24 <kaniini> mitchty: why disable PIE?
2015-06-18 00:24:52 <kaniini> mitchty: why also paxctl?   if anything, paxmark should now be used instead, but, why?
2015-06-18 00:26:10 <kaniini> mitchty: beyond that, the use of docker to build a bootstrap ghc seems unnecessary to me
2015-06-18 00:26:29 <kaniini> mitchty: maybe for the *first* version but once we have an adequate host-ghc we can do it from alpine itself
2015-06-18 03:38:49 <mitchty> kaniini so first bit, without disabling pie this happens https://gist.github.com/mitchty/29125de21ade1a04364f
2015-06-18 03:39:18 <mitchty> i've managed to build simple programs without -nopie, but i cannot build cabal, or ghc itself that way
2015-06-18 03:39:34 <mitchty> i've not looked into it a ton however
2015-06-18 03:40:13 <mitchty> as for paxctl, went for the first thing I found that worked for building the helloworld example in the ghc source tree
2015-06-18 03:41:09 <mitchty> as a note, even with a native host compiling wihtout stage1/2, you'l end up with ld trying to link non PIC with PIC in shared libraries, or seemingly trying to
2015-06-18 03:41:53 <mitchty> you can force ghc to use PIC in other ways for shared libraries but not executables but i'm not 100% on how PaX impacts this
2015-06-18 03:42:50 <mitchty> but past that i agree its an OK first pass to get a native compiler onto alpine and iterate from there, as a note even with -nopie c ffi works fine from my limited testing
2015-06-18 03:44:01 <mitchty> i've been comparing the ports of ghc in gentoo with hardening and debian as well as fedora, not entirely sure that I know why -nopie is needed yet
2015-06-18 03:44:18 <mitchty> and lastly only used docker because its convenient enough
2015-06-18 04:47:30 <kaniini> mitchty: my concern with -nopie is the security impact
2015-06-18 04:49:41 <kaniini> mitchty: i am confused as to why a program written in a functional language would at all be affected by ASLR
2015-06-18 04:54:38 <mitchty> kaniini well if you look at the gist the issue seems to reside in the runtime system, so i'm going to guess it might be related how haskell deals with lazy evaluation of thunks, but i'm a bit confused why this only shows up on alpine linux and not gentoo when compiled with musl. which mostly leads me to think its the security patches which I admit I have only begun to review
2015-06-18 04:55:06 <mitchty> i normally work with kernel modules in linux so this user side of things is fun but a bit maddening
2015-06-18 04:56:02 <mitchty> i did a quick objdump on that archive and don't quite understand why ld would complain here
2015-06-18 04:57:00 <mitchty> i didn't see non PIC code in it but hadn't yet looked to see what patches might have been applied to gcc
2015-06-18 04:57:31 <mitchty> but thanks for the quick review in either case!
2015-06-18 05:04:02 <mitchty> as a note, part of the reason for -nopie is I had segfaults in the runtime when compiling itself, and https://ghc.haskell.org/trac/ghc/wiki/Debugging/CompiledCode recommended disabling aslr while debugging
2015-06-18 05:08:42 <mitchty> a quick look a a few old bugs regarding aslr reads that the garbage collector might not like it, but i might have ported this poorly as osx 10.7 went fully PIE as well so possibly a build issue too
2015-06-18 07:47:34 <ScrumpyJack> I vote for the commits in this channel. I already have too many channels open :)
2015-06-18 09:04:33 <clandmeter> this channel is awfull quiet wihtout algitbot spamming us
2015-06-18 09:04:47 <clandmeter> doesnt feel lik home :)
2015-06-18 09:14:33 <ncopa> i wonder if we want all commits
2015-06-18 09:14:38 <ncopa> or only when there is a push
2015-06-18 09:14:49 <ncopa> you can push 10 commits in one git push
2015-06-18 09:15:08 <barthalion> I don't mind having commit messages on different channel
2015-06-18 09:15:28 <ncopa> i think i'll just post up to 20 commits in one push
2015-06-18 09:15:45 <ncopa> the idea is to publish the push on mqtt as json
2015-06-18 09:16:27 <clandmeter> my doubht is, when somebody requests something here, its nice to have a commit to show the result.
2015-06-18 09:16:43 <ncopa> thats true
2015-06-18 09:16:45 <clandmeter> but then again, spamming in discussions is also bad.
2015-06-18 09:16:58 <clandmeter> so im not convinced about any solution yet.
2015-06-18 09:17:09 <barthalion> weechat can merge buffers if you wish to
2015-06-18 09:31:53 <clandmeter> im slightly in favor of having it in alpine-aports. it can also be more verbose, and maybe add some switches to our bot to ask for build status?
2015-06-18 09:35:20 <clandmeter> ncopa: does cgit support commit range?
2015-06-18 09:36:12 <barthalion> also, where is retry command to algitbot :(
2015-06-18 09:36:52 <clandmeter> +1
2015-06-18 09:37:13 <ncopa> i'll fix those too soonish
2015-06-18 09:37:22 <ncopa> i'm writing new git push -> mqtt script
2015-06-18 09:37:24 <ncopa> in lua
2015-06-18 09:37:51 <ncopa> i think i'll have the topic: git/<repo>/commits
2015-06-18 09:38:00 <ncopa> where each commit is pushed in json format
2015-06-18 09:38:22 <barthalion> we need to go deeper; let's post new patchwork entries on #a-aports too
2015-06-18 09:38:36 <ncopa> i am not sure how to do the <repo> thing though
2015-06-18 09:39:05 <ncopa> for example: user/ncopa/testrepo
2015-06-18 09:39:15 <ncopa> or if i do: user:ncopa:testreop
2015-06-18 09:39:51 <ncopa> this is due to the mqtt topic: git/user/ncopa/testrepo/commits vs git/user:ncopa:testrepo/commits
2015-06-18 09:40:06 <ncopa> i dont know what we prefer
2015-06-18 09:40:23 <ncopa> i suppose the former is better
2015-06-18 09:40:38 <ncopa> since then you can subscribe to a given user only if you want
2015-06-18 09:41:06 <clandmeter> what if i dont want my commits to be pushed to msg?
2015-06-18 09:41:27 <ncopa> i think thats not possible atm
2015-06-18 09:41:48 <ncopa> then we probably need to do per repo hooks
2015-06-18 09:41:48 <clandmeter> i mean for non aports
2015-06-18 09:42:28 <ncopa> currently i have installed the post-receive hook on all git repos
2015-06-18 09:42:31 <clandmeter> although i dont know any reason to not want it.
2015-06-18 09:43:03 <ncopa> i also need a script that generate tarballs when someone pushes git tag
2015-06-18 09:43:28 <clandmeter> push them where?
2015-06-18 09:43:51 <ncopa> git push --tags
2015-06-18 09:43:55 <clandmeter> the tarballs on cgit are not enough?
2015-06-18 09:43:57 <ncopa> push the tag to git.a.o
2015-06-18 09:43:58 <ncopa> no
2015-06-18 09:44:02 <ncopa> thats the problem
2015-06-18 09:44:04 <clandmeter> chechsum?
2015-06-18 09:44:07 <ncopa> yes
2015-06-18 09:44:12 <ncopa> they are generated on the fly
2015-06-18 09:44:31 <clandmeter> cgit should cache them like github does
2015-06-18 09:44:31 <ScrumpyJack> I'm getting ERROR package a trying to overwrite /some/file owned by b
2015-06-18 09:44:53 <ncopa> so if we update cgit, it might generate the tarball differently
2015-06-18 09:44:57 <ncopa> and checksum fails
2015-06-18 09:45:10 <ncopa> ScrumpyJack: which package and which file?
2015-06-18 09:45:13 <clandmeter> not if we patch cgit :)
2015-06-18 09:45:27 <ncopa> clandmeter: i actually though of that
2015-06-18 09:45:38 <ScrumpyJack> stuff in /usr/share/mime
2015-06-18 09:45:45 <barthalion> it's not true anymore, guys
2015-06-18 09:45:54 <ncopa> clandmeter: i'd like cgit cache the tarball
2015-06-18 09:46:00 <ScrumpyJack> the packages are mate-control-center and bluefish
2015-06-18 09:46:01 <barthalion> even if cgit still generates them on the fly, checksum doesn't change anymore
2015-06-18 09:46:12 <barthalion> it has been fixed quite a long time ago
2015-06-18 09:46:24 <ncopa> barthalion: they fixed the gz timestamp issue
2015-06-18 09:46:39 <barthalion> sounds like good enough?
2015-06-18 09:46:42 <clandmeter> i think i had that issue rather recently
2015-06-18 09:47:00 <barthalion> well, I had not
2015-06-18 09:47:08 <clandmeter> hmm
2015-06-18 09:47:15 <ncopa> barthalion: i have noticed that if you update cgit, and cgit uses newer git version under the hood, then it might change how the tarball is generated
2015-06-18 09:47:18 <clandmeter> maybe somebody used an older version of cgit
2015-06-18 09:47:46 <ncopa> even the order on how things are stored
2015-06-18 09:47:50 <ncopa> or the internal tar fields
2015-06-18 09:47:52 <barthalion> ncopa: maybe; but upstream is generally helpful, so why we don't throw our problems at them?
2015-06-18 09:47:59 <ncopa> yup
2015-06-18 09:48:14 <ncopa> feature request for upstream: cache generated tarballs
2015-06-18 09:48:51 <ncopa> i talked with them yesterday on irc about other problem
2015-06-18 09:48:57 <ncopa> but didnt get any response
2015-06-18 09:49:01 <ncopa> problem is clone url
2015-06-18 09:49:19 <ncopa> clone-prefix=git@git.alpinelinux.org:
2015-06-18 09:49:58 <ncopa> will display the clone url: git@git.alpinelinux.org:/repo
2015-06-18 09:50:09 <ncopa> the leading / infront of repo shouldnt be there
2015-06-18 09:50:25 <ncopa> http://sprunge.us/UcZA
2015-06-18 09:50:42 <ncopa> i need to test it though...
2015-06-18 09:54:11 <clandmeter> ncopa: so which road are we going to travel? a full fledged alpine-aports channel?
2015-06-18 09:54:37 <ncopa> dont know yet
2015-06-18 09:54:46 <ncopa> i think i fix the git hook first
2015-06-18 09:54:59 <ncopa> and then deal with which irc channel
2015-06-18 09:55:03 <clandmeter> :)
2015-06-18 09:55:17 <clandmeter> you shouldnt set priorities, its no fun.
2015-06-18 09:55:26 <ncopa> :)
2015-06-18 09:55:41 <ncopa> but really
2015-06-18 09:55:52 <ncopa> #alpine-aports sounds like aports only
2015-06-18 09:55:58 <ncopa> we could have #alpine-commits
2015-06-18 09:56:05 <ncopa> and you see all commits in all git repos
2015-06-18 09:56:28 <clandmeter> hmm
2015-06-18 09:56:49 <ncopa> so you also see commits in abuild, apk-tools etc
2015-06-18 09:57:12 <ncopa> i am slightly in favor of using separate channel for commits
2015-06-18 09:57:27 <ncopa> it tend to be noisy in here
2015-06-18 09:57:53 <barthalion> #alpine-spam
2015-06-18 09:57:56 <barthalion> #alpine-noise
2015-06-18 09:57:58 <barthalion> #alpine-bots
2015-06-18 09:58:09 <clandmeter> ok, i can always ignore other project commits if i dont like them.
2015-06-18 09:58:37 <clandmeter> #alpine-twitter
2015-06-18 09:58:52 <fcolista> I like alpine-twitter :)
2015-06-18 09:59:46 <clandmeter> and we can have colors in #alpine-whatever :)
2015-06-18 10:01:17 <clandmeter> barthalion: that patchwork idea, is it on the same list as your previous suggestion? ;-)
2015-06-18 10:01:35 <ncopa> btw, there is a http://git.alpinelinux.org/cgit/user/sl/build-server-status/
2015-06-18 10:01:42 <ncopa> which shows what build servers are doing
2015-06-18 10:01:51 <ncopa> mqtt websockets
2015-06-18 10:01:54 <ncopa> pretty nice
2015-06-18 10:02:08 <barthalion> clandmeter: which one? I make a lot of suggestions that I never implement 8)
2015-06-18 10:02:36 <clandmeter> yes that/those one(s) :P
2015-06-18 10:03:08 <barthalion> I promise I'll clone myself as soon as it'll be possible!
2015-06-18 10:03:13 <barthalion> but we can just clone ncopa then
2015-06-18 10:03:31 <ncopa> :)
2015-06-18 10:03:38 <ncopa> sounds like i better have a lunch...
2015-06-18 10:04:20 <vkris> is there any policy from irc people on bot msg infested channel ?
2015-06-18 10:09:14 <clandmeter> vkris: ofcourse, no cursing except for me.
2015-06-18 10:09:55 <vkris> maybe then #alpine-builders
2015-06-18 10:10:20 <vkris> and enhanced algibot commands for more querying
2015-06-18 10:10:47 <clandmeter> if we merge all those channel names together, we will end up with alpine-devel
2015-06-18 10:11:05 <vkris> to avoid overloading commands on servers redis-cached msg
2015-06-18 10:11:27 <vkris> but we want bot msgs seperate
2015-06-18 10:13:07 <ScrumpyJack> can apk take input from a file?
2015-06-18 10:13:58 <vkris> there could be some algibot command like  /command redir:devel, to send to -devel channel
2015-06-18 10:13:58 <ScrumpyJack> i mean apk add
2015-06-18 10:14:23 <vkris> that way both channels are happy
2015-06-18 10:15:01 <vkris> ScrumpyJack: update gammu pls
2015-06-18 10:15:16 <vkris> lots news versions
2015-06-18 10:15:35 <vkris> and there is now a nice python binding too
2015-06-18 10:15:36 <ScrumpyJack> vkris: okey dokey
2015-06-18 10:15:46 <vkris> thanks
2015-06-18 10:23:11 <vkris> all msgs needs to numbered
2015-06-18 10:23:34 <vkris> so "algibot: redir <num> -devel"
2015-06-18 10:24:19 <vkris> that would fetch <num> from redis
2015-06-18 10:26:25 <vkris> that way only core developers need to be on -builders channel
2015-06-18 11:03:56 <ScrumpyJack> ncopa: could you bump i3wm to 4.10?
2015-06-18 11:04:20 <ncopa> sure
2015-06-18 11:06:50 <ncopa> done
2015-06-18 11:19:21 <ScrumpyJack> thank you very much
2015-06-18 12:05:08 <mosez> hum... how to i update a local user? no usermod?
2015-06-18 12:06:16 <barthalion> usermod probably comes from util-linux
2015-06-18 12:09:58 <mosez> nope :(
2015-06-18 12:10:36 <barthalion> hum
2015-06-18 12:11:05 <clandmeter> leo-unglaub: /names
2015-06-18 12:11:09 <clandmeter> grr
2015-06-18 12:11:35 <barthalion> mosez: no idea then, you can edit /etc/passwd or /etc/group manually anyway
2015-06-18 12:37:21 <ahills> moduse
2015-06-18 12:37:27 <ahills> r?
2015-06-18 12:37:32 <ahills> Stupid phone
2015-06-18 12:38:13 <ahills> ncopa: for the record I'm 100% in favor of a separate channel #alpine-commits
2015-06-18 12:38:22 <mosez> # mod
2015-06-18 12:38:23 <mosez> modinfo   modprobe
2015-06-18 12:38:39 <ahills> Also do you actually like using lua?
2015-06-18 12:39:05 <mosez> ahills: the main alpine guys fall in love with lua :D
2015-06-18 12:39:11 <mosez> especially turbolua
2015-06-18 12:39:36 <ScrumpyJack> do we have libopenjpeg anywhere?
2015-06-18 12:41:16 <ahills> I like many of the ideas behind lua and also the size but I hate being deprived of very basic language features like +=
2015-06-18 12:41:28 <clandmeter> ScrumpyJack: http://pkgs.alpinelinux.org/packages?package=%25jpeg%25&repo=all&arch=x86
2015-06-18 12:41:56 <ahills> It makes trivial tasks rather tedious
2015-06-18 12:41:57 <clandmeter> talking about turbolua... :)
2015-06-18 12:43:17 <ahills> It just makes me want lisp more
2015-06-18 12:43:38 <ScrumpyJack> ncopa: would you mind awfully rebuilding encfs? (quite a lot of stuff is broken in edge at the moment :)
2015-06-18 12:44:10 <ScrumpyJack> clandmeter: none of those provides openlibjpeg1, as far as I can tell :(
2015-06-18 12:44:20 <ScrumpyJack> or rather lobopenjpeg1 :)
2015-06-18 12:44:24 <ScrumpyJack> lib, even
2015-06-18 12:44:27 <ScrumpyJack> sheesh!
2015-06-18 12:44:40 <ncopa> does rlog need to be rebuilt too?
2015-06-18 12:44:50 <ScrumpyJack> probably :)
2015-06-18 12:45:30 <clandmeter> ScrumpyJack: probably disabled in openjpeg?
2015-06-18 12:48:19 <ahills> ScrumpyJack: usually the symlink for /\.so$/ comes from the -dev package
2015-06-18 12:48:26 <clandmeter> ScrumpyJack: v1 is pretty old isnt it?
2015-06-18 12:48:36 <clandmeter> arch still has v1
2015-06-18 12:51:17 <barthalion> we have both
2015-06-18 12:51:57 <barthalion> almost nothing uses v2, and API changes were too big to care
2015-06-18 12:52:20 <ScrumpyJack> barthalion: have you found what provides libopenjpeg1?
2015-06-18 12:52:31 <barthalion> in Alpine? probably nothing
2015-06-18 12:52:56 <barthalion> I was talking about Arch
2015-06-18 12:53:09 <ScrumpyJack> ah! heh :)
2015-06-18 12:53:09 <clandmeter> we would need openjpeg1
2015-06-18 13:05:21 <clandmeter> ScrumpyJack: its in edge/testing now
2015-06-18 13:19:15 <ScrumpyJack> clandmeter: thank you
2015-06-18 13:19:22 <ncopa> hum
2015-06-18 13:26:05 <ncopa> ok mqtt hooks now use json as payload data
2015-06-18 13:39:38 <alacerda> please: http://sprunge.us/cUfU
2015-06-18 13:40:37 <ncopa> done
2015-06-18 13:40:55 <alacerda> thanks
2015-06-18 13:41:13 <BitL0G1c> mosez / bathalion - usermod is in shadow http://pkgs.alpinelinux.org/contents?pkgname=shadow&arch=x86
2015-06-18 13:47:25 <mosez> bitl0g1c: and shadow is in testing.
2015-06-18 13:50:45 <ahills> mosez: you can also vi /etc/passwd
2015-06-18 14:02:01 <BitL0G1c> yes in testing
2015-06-18 14:35:18 <ScrumpyJack> ncopa: could you rebuild libexr ?
2015-06-18 14:36:11 <ncopa> cannot find libexr
2015-06-18 14:46:26 <ncopa> do we want algitbot post messages here when new packages are uploaded?
2015-06-18 14:46:39 <ncopa> or do we publish those on #alpine-commits too?
2015-06-18 14:48:44 <fcolista> if packages are posted here,  think there are no reasons for #alpine-commits
2015-06-18 14:49:08 <fcolista> maybe the existence of #alpine-commit could be added in the topic?
2015-06-18 14:49:11 <ncopa> commits are posted in #alpine-commits
2015-06-18 14:49:23 <ncopa> all commits
2015-06-18 14:49:50 <fcolista> oh, you mean just new packages here ?
2015-06-18 14:49:55 <ncopa> like:
2015-06-18 14:50:12 <ncopa> like those^^^
2015-06-18 14:52:09 <fcolista> ScrumpyJack, re: ghtumb
2015-06-18 14:52:19 <fcolista> can I add it as maintainer?
2015-06-18 14:52:28 <fcolista> *can I add you
2015-06-18 14:53:44 <ScrumpyJack> yes, ok
2015-06-18 14:54:18 <ScrumpyJack> i am not down as maintainer on any package that i have contributed :(
2015-06-18 14:54:55 <fcolista> ok.
2015-06-18 14:56:54 <alacerda> uia
2015-06-18 14:57:05 <alacerda> ops... whong window, sorry
2015-06-18 14:57:09 <alacerda> wrong*
2015-06-18 19:14:13 <jomat> d
2015-06-18 19:14:57 <jomat> sorry... that should have been a ctrl-b d return ~. to exit tmux and ssh :-)
2015-06-18 19:16:38 <jomat> btw: greetings from a warmup party for hackmeeting.org :-)
2015-06-18 19:27:07 <alacerda> jomat, _o/
2015-06-19 06:23:19 <barthalion> ncopa: 'uploaded' packages should also go to -commits
2015-06-19 06:23:30 <ncopa> ok
2015-06-19 06:23:48 <barthalion> is it possible to have only "failed" messages here?
2015-06-19 06:24:01 <ncopa> yes
2015-06-19 06:24:06 <ncopa> those need a bit work
2015-06-19 06:24:15 <ncopa> i need to collect the build logs some place
2015-06-19 06:24:18 <barthalion> so I think it makes sense to have only failures here
2015-06-19 06:24:36 <barthalion> because this is more important to have working CI
2015-06-19 06:24:59 <ncopa> i.e when build is done, then scp the build logs to some central place
2015-06-19 06:25:19 <ncopa> so we can publish exact url with error message
2015-06-19 07:28:28 <ScrumpyJack> are commits in mqtt?
2015-06-19 08:08:24 <ScrumpyJack> building esd, i get ltconfig: you must specify a host type if you use `--no-verify'
2015-06-19 08:09:40 <ScrumpyJack> i have specified --build=$CBUILD and --host=$CHOST which isn't recognised
2015-06-19 08:11:17 <ScrumpyJack> this is during configure
2015-06-19 08:12:00 <ScrumpyJack> removing --no-verify from the configure script makes things worse :)
2015-06-19 08:24:41 <ScrumpyJack> there isn't much out there aboud esd. it's unmaintained, but i need a package on Alpine to support a sound server for legacy systems that have esd. Has anyone seen this issue with libtool before?
2015-06-19 08:31:17 <ScrumpyJack> hmm, ltconfig is nowhere to be seen
2015-06-19 08:41:31 <ScrumpyJack> removed the offending libtool configuration
2015-06-19 09:06:00 <pluesch> ScrumpyJack: any news about opensmtpd and musl/alpinelinux?
2015-06-19 09:06:41 <pluesch> saw your bug report on bugs.a.o
2015-06-19 09:09:23 <barthalion> what's about it? can you link it?
2015-06-19 09:35:06 <pluesch> barthalion: sure thing. http://bugs.alpinelinux.org/issues/4012
2015-06-19 09:36:17 <barthalion> clandmeter, ncopa: can you reset my password?
2015-06-19 09:36:27 <barthalion> I mean to redmine
2015-06-19 09:36:39 <barthalion> pluesch: I'll take care of it, I need to review mongodb patches first
2015-06-19 09:37:23 <ncopa> clandmeter: do you help barthalion?
2015-06-19 09:37:47 <pluesch> barthalion: hah. nice! thank you :3
2015-06-19 09:37:58 <ncopa> clandmeter: nm. im on it
2015-06-19 09:43:25 <clandmeter> did i mis someting?
2015-06-19 09:45:54 <barthalion> no, we're cool
2015-06-19 10:06:28 <ScrumpyJack> barthalion: you having a go at opensmtpd?
2015-06-19 10:07:55 <barthalion> ScrumpyJack: I'll be
2015-06-19 10:08:07 <barthalion> mongodb first
2015-06-19 10:08:11 <barthalion> I promised it long time ago
2015-06-19 10:10:21 <barthalion> but that after 1PM, I have a meeting in a few minutes
2015-06-19 10:10:39 <barthalion> so, assuming I don't screw up my time management again, I'll push opensmtpd this weekend
2015-06-19 10:11:21 <ScrumpyJack> ok. no problem. thnaks for looking at it. from my little experiance, you'll need libasr, which builds fine an AL, but then opensmtpd won't build against it. it's a known problem apparently.
2015-06-19 10:12:11 <barthalion> I do puppet for a living
2015-06-19 10:12:17 <barthalion> no compilation error can stop me
2015-06-19 10:12:23 <ScrumpyJack> lol :)
2015-06-19 10:12:59 <ScrumpyJack> you need a car sticker that says "I stop for no compilation error"
2015-06-19 10:27:20 <ScrumpyJack> I don't know much about icons. how do they work in a desktop env? does someone create a whole bunch of icons for every app under the sun, package the up for a distro so they go in /usr/share/icons or something?
2015-06-19 10:28:11 <barthalion> for every filetype
2015-06-19 10:28:13 <barthalion> pretty much yes
2015-06-19 10:29:10 <ScrumpyJack> we seem to have a faenza-icon-theme. mate has a theme. are these themes centralised somewhere on the intertubes?
2015-06-19 10:29:44 <barthalion> not sure if I understand now :(
2015-06-19 10:29:48 <barthalion> what is centralised?
2015-06-19 10:30:01 <barthalion> there are some common elements, usually provided by gnome-icon-theme or sth like that
2015-06-19 10:35:40 <ScrumpyJack> centralised as in, somewhere easy to find them all
2015-06-19 10:36:19 <barthalion> gnome-look.org, most likely
2015-06-19 10:36:24 <barthalion> and deviantart
2015-06-19 11:11:11 <clandmeter> barthalion: im looking into opensmtpd. it builds but linker is not working.
2015-06-19 11:11:44 <clandmeter> i asked ncopa to look into it, cause i couldnt find the issue.
2015-06-19 11:24:28 <ncopa> will look at it soonish
2015-06-19 12:55:53 <ncopa> algitbot: hi
2015-06-19 13:04:54 <ncopa> uploads are no longer echoed here
2015-06-19 13:04:57 <ncopa> hum
2015-06-19 13:05:04 <ncopa> in fact, not errors either
2015-06-19 13:22:51 <fcolista> wb algitbot
2015-06-19 13:29:47 <shafire> hi
2015-06-19 13:30:01 <ncopa> hi
2015-06-19 13:38:54 <ncopa> algitbot: retry master
2015-06-19 13:42:10 <ncopa> wors too
2015-06-19 13:42:14 <ncopa> works*
2015-06-19 13:42:21 <barthalion> why no messages then?
2015-06-19 13:42:26 <barthalion> because there is nothing to retry?
2015-06-19 13:42:29 <ncopa> because there are nothing to do
2015-06-19 13:42:33 <barthalion> k
2015-06-19 13:42:40 <ncopa> there are no new commits
2015-06-19 13:42:54 <ncopa> the alpine-commits only publish new commits
2015-06-19 13:43:07 <ncopa> do we want more verbose messages in #alpine-commits?
2015-06-19 13:43:20 <barthalion> depends what more verbosity means
2015-06-19 13:43:52 <vkris> http://git.alpinelinux.org/cgit/sircbot/ <-- No repositories found
2015-06-19 13:44:20 <vkris> is it ok to pkg bin in http://pkgs.alpinelinux.org/contents?pkgname=libxslt&arch=x86 , has /usr/bin/xsltproc ?
2015-06-19 13:46:22 <barthalion> vkris: no
2015-06-19 13:46:29 <barthalion> remind me folks how to subscribe to our mqtt
2015-06-19 13:47:48 <vkris> then maybe pkg xsltproc
2015-06-19 13:48:55 <alacerda> vkris, were you looking for this: http://git.alpinelinux.org/cgit/hosted/sircbot/ ?
2015-06-19 13:49:17 <vkris> yes, thanks
2015-06-19 13:49:48 <vkris> may fix the url here, http://git.alpinelinux.org/cgit/aports/commit/?id=c01343bad073860e4d0392bdde6623bf0e3678a8
2015-06-19 13:52:14 <shafire> I asked for a zfs iso flavor :)
2015-06-19 13:55:40 <clandmeter> shafire: i have an iso on my dev page
2015-06-19 13:59:13 <clandmeter> but its a few months old.
2015-06-19 14:05:24 <ncopa> clandmeter: this? /usr/lib/gcc/x86_64-alpine-linux-musl/5.1.0/../../../../lib/libasr.so: undefined reference to `res_randomid'
2015-06-19 14:06:30 <shafire> clandmeter: I see
2015-06-19 14:07:03 <shafire> I just want to have the iso also with the normal releases
2015-06-19 14:09:03 <ncopa> clandmeter: libasr assmues the libc has res_randomid()
2015-06-19 14:09:11 <ncopa> its a bug in libasr
2015-06-19 14:09:33 <ncopa> there should probably be a configure check and use a fallback if its missing
2015-06-19 14:10:29 <ncopa> ha
2015-06-19 14:11:04 <ncopa> 0573 u_int
2015-06-19 14:11:04 <ncopa> 0574 res_randomid(void) {
2015-06-19 14:11:04 <ncopa> 0575     return 0xffff & __getpid();
2015-06-19 14:11:04 <ncopa> 0576 }
2015-06-19 14:11:13 <ncopa> so much random...
2015-06-19 14:14:13 <ncopa> clandmeter: this is your issue: https://github.com/OpenSMTPD/libasr/issues/8
2015-06-19 14:29:34 <ncopa> clandmeter: this might work around it for how: https://github.com/OpenSMTPD/libasr/issues/8#issuecomment-113531155
2015-06-19 14:31:45 <ncopa> have a nice weekend
2015-06-19 14:50:21 <stateless> ncopa, did you intend to write 0xfff
2015-06-19 14:50:26 <stateless> not that it makes a difference given how broken it is
2015-06-19 14:50:27 <stateless> :P
2015-06-19 15:33:25 <pluesch> -.- why is this happening...
2015-06-19 15:52:43 <alacerda> pluesch, what?
2015-06-19 15:56:23 <pluesch> res_randomid glibc implementation
2015-06-19 15:56:30 <pluesch> return 0xffff & __getpid();
2015-06-19 17:43:35 <clandmeter> somebody rsyncing on rsync.a.o?
2015-06-19 19:58:52 <elkw> Hello
2015-06-19 19:58:53 <elkw> I accidentally purged musl and many other packages while trying to downgrade musl
2015-06-19 19:59:02 <elkw> now my computer won`t boot and I`m trying to fix it
2015-06-19 19:59:09 <elkw> I`m on a live cd, mounted my hd, copied busybox-static and apk-static to it
2015-06-19 19:59:15 <elkw> and then did a chroot
2015-06-19 19:59:24 <elkw> now I`m trying to run apk-static and add musl back
2015-06-19 19:59:28 <elkw> however...
2015-06-19 19:59:34 <elkw> # /home/apk.static update
2015-06-19 19:59:38 <elkw> ERROR: Unable to lock database: No such file or directory
2015-06-19 19:59:44 <elkw> ERROR: Failed to open apk database: No such file or directory
2015-06-19 19:59:51 <elkw> I have internet working, according to ping 8.8.8.8 so that`s not the issue
2015-06-19 20:00:00 <elkw> Any idea on how to fix this? It`s the last step preventing me from fixing my computer (or so I hope)
2015-06-19 22:02:50 <systmkor> elkw: use apk.static to reinstall it
2015-06-19 22:03:23 <systmkor> "/sbin/apk.static"
2015-06-19 22:03:52 <systmkor> assuming to execute that you don't need musl-libc
2015-06-19 22:05:19 <clandmeter> he is trying :)
2015-06-19 22:08:46 <elkw> systmkor: thanks. clandmeter told me that on a PM
2015-06-19 22:09:52 <elkw> I`m trying, but this is some tricky stuff...
2015-06-19 22:11:07 <clandmeter> he managed to remove musl and his kernel and 450 other packages
2015-06-19 22:12:54 <systmkor> clandmeter: have you ran the alpine generic arm release?
2015-06-19 22:13:13 <clandmeter> the uboot one?
2015-06-19 22:24:02 <systmkor> yah
2015-06-19 22:24:21 <systmkor> clandmeter: yes, the generic one I think it's called armhf instead of arm-rpi
2015-06-19 22:41:41 <clandmeter> i only used the rpi one
2015-06-19 22:46:45 <systmkor> ah dang, been trying to get the arm generic one running on qemu and it just keeps crashing on me
2015-06-19 22:47:06 <systmkor> i didn't find any alpine wiki articles for it
2015-06-19 23:22:33 <elkw> okay, got everything installed again, but it still won`t boot
2015-06-19 23:22:39 <elkw> something`s wrong with the kernel
2015-06-19 23:22:50 <elkw> when I did apk add linux-grsec I got
2015-06-19 23:22:57 <elkw> depmod: can`t change directory to 'lib/modules/3.18.14-1-grsec': No such file or directory
2015-06-19 23:23:03 <elkw> extlinux: cannot find device got path /boot
2015-06-19 23:23:08 <elkw> extlinux: cannot open device (null)
2015-06-19 23:23:55 <elkw> I ignored and tried to boot. It said: Kernel panic - not syncing: No init found. Try passing init = option to kernel
2015-06-19 23:24:15 <elkw> what should I do? I`m back at the live cd now
2015-06-19 23:25:38 <elkw> now I del and add syslinux
2015-06-19 23:25:40 <ahills> elkw: update-extlinux takes a kernel version argument
2015-06-19 23:25:47 <elkw> same error as with linux-grsec
2015-06-19 23:25:53 <elkw> extlinux: cannot find device got path /boot
2015-06-19 23:25:56 <elkw> extlinux: cannot open device (null)
2015-06-19 23:26:40 <elkw> ahills: what should I do then? I`m on a live cd with apk-static (because apk doesn`t work inside chroot)
2015-06-19 23:27:01 <elkw> I mounted /dev/sda3 on /mnt and /dev/sda1 on /mnt/boot
2015-06-19 23:28:13 <ahills> update-extlinux -h and also read the .conf
2015-06-19 23:28:23 <ahills> It's pretty short
2015-06-19 23:29:05 <ahills> Look at your /mnt/lib
2015-06-19 23:29:21 <elkw> will try, thanks
2015-06-19 23:29:30 <elkw> but I do that inside chroot, right?
2015-06-19 23:29:33 <ahills> The folder in there is the exact version
2015-06-19 23:29:37 <ahills> Yes
2015-06-19 23:29:58 <ahills> You could do it outside, which is what I do
2015-06-19 23:30:26 <elkw> outside I don`t have the update-extlinux command
2015-06-19 23:30:34 <ahills> I apk update/upgrade the kernel then run it with the right flags
2015-06-19 23:30:45 <ahills> Install it then :P
2015-06-19 23:30:56 <ahills> It's a complete Alpine system...
2015-06-19 23:31:04 <elkw> apk doesn`t work inside chroot
2015-06-19 23:31:17 <ahills> No. Outside,
2015-06-19 23:31:32 <ahills> It can all be done without chroot
2015-06-19 23:31:41 <elkw> nice
2015-06-19 23:32:11 <ahills> You just have to remember to always give prefix arguments for commands that have them
2015-06-19 23:36:14 <elkw> # /sbin/update-extlinux -h
2015-06-19 23:36:54 <elkw> usage: /sbin/update-extlinux [v|--verbose] [warn-only]
2015-06-19 23:37:04 <elkw> doesn`t talk about kernel version arguments
2015-06-19 23:38:34 <ahills> It's a shell script so you can probably figure out where it's failing
2015-06-19 23:39:00 <ahills> Use sh -x if you can't just via the error
2015-06-19 23:39:18 <ahills> I have to catch a train, sorry
2015-06-19 23:39:27 <elkw> don`t worry, I`ll keep trying
2015-06-19 23:39:29 <elkw> thanks a lot
2015-06-19 23:39:31 <elkw> have a nice trip
2015-06-20 00:21:43 <elkw> I just noticed my /etc/mtab file is empty!
2015-06-20 00:21:52 <elkw> maybe that`s the reason for the errors?
2015-06-20 00:30:57 <elkw> And that was it indeed!
2015-06-20 00:31:14 <elkw> finally got my computer to boot again
2015-06-20 00:31:32 <elkw> Tip of the day: do NOT try to downgrade musl
2015-06-20 00:31:47 <elkw> Bug report: apk doesn`t work inside chroot
2015-06-20 07:38:57 <ttilley> ncopa, i finally got off my lazy ass and did what you suggested for clang
2015-06-20 07:39:09 <ttilley> ncopa, http://bugs.alpinelinux.org/issues/4374
2015-06-20 07:40:29 <ttilley> it now does -Wl,-z,nwo -Wl,-z,relro by default and enables SSP by default (it already had PIE by default)
2015-06-20 07:41:02 <ttilley> also added polly support now that isl 0.14 is available
2015-06-20 07:41:40 <ttilley> polly doesn't have a sane fail-back if the isl it wants isn't there, however. it just won't work.
2015-06-20 07:42:56 <ttilley> 
ACTION thinks it's pretty neat that gcc can do that

2015-06-20 15:28:26 <shafire> hi
2015-06-21 09:57:32 <stateless> ncopa, dalias found a few problems with fortify, will push out some fixes next week and let you know
2015-06-22 08:13:52 <fcolista> ncopa, barthalion: re %212, please remember me how do we want to manage messages after install. iirc having  them in post-install (especially with coloured ouput) was no good.
2015-06-22 08:14:11 <fcolista> http://sprunge.us/XQTj
2015-06-22 08:14:18 <fcolista> that's the .post-install
2015-06-22 08:14:54 <ncopa> we dont want that
2015-06-22 08:16:10 <ncopa> i think we can just remove the post-install
2015-06-22 08:16:35 <fcolista> ok
2015-06-22 08:16:47 <fcolista> hwo do we manage those messages?
2015-06-22 08:17:09 <ncopa> currently we dont
2015-06-22 08:17:20 <fcolista> ok
2015-06-22 08:17:28 <ncopa> i think there was something talk about adding them to a /usr/share/doc/$pkgname/README.alpine
2015-06-22 08:17:40 <fcolista> yes, i remember something like that
2015-06-22 08:23:39 <clandmeter> ncopa: does apk suppress those msgs with --quiet?
2015-06-22 08:24:55 <ncopa> clandmeter: which messages?
2015-06-22 08:25:00 <clandmeter> if not maybe have --really-quiet :)
2015-06-22 08:25:01 <ncopa> messages from .post-install scripts?
2015-06-22 08:25:04 <ncopa> no
2015-06-22 08:25:04 <clandmeter> yeah
2015-06-22 08:25:45 <ncopa> afaik apk does not try do do anything with anything from *install scripts
2015-06-22 08:25:51 <ncopa> it just executes them
2015-06-22 08:25:52 <ncopa> thats it
2015-06-22 08:26:10 <clandmeter> ok, but the main issue with those msg are when running from tmpfs right?
2015-06-22 08:26:46 <ncopa> the main issue with those messages are that they are making assumptions
2015-06-22 08:26:55 <ncopa> they assme that 1: someone is actually reading the output
2015-06-22 08:27:13 <ncopa> 2: that the stdout is actually a terminal
2015-06-22 08:27:49 <ncopa> 3: they add additional fork/exec -> slows things down
2015-06-22 08:28:03 <ncopa> it does not matter if a single package has a .post-install
2015-06-22 08:28:20 <ncopa> but if every single package had it, things would be noticable slower
2015-06-22 08:28:46 <ncopa> triggers are much better for this kind of things
2015-06-22 08:29:02 <ncopa> as they are executed after all packages are installed
2015-06-22 08:32:01 <ncopa> example for problem 1: apk add <list with 300 packages>
2015-06-22 08:32:13 <ncopa> do you think the message will be read?
2015-06-22 08:32:22 <barthalion> it won't
2015-06-22 08:32:28 <barthalion> source: we do it in Arch
2015-06-22 08:32:34 <barthalion> I constantly miss these messages
2015-06-22 08:32:41 <ncopa> example for problem 2:
2015-06-22 08:32:43 <clandmeter> triggers seems nice
2015-06-22 08:33:22 <ncopa> in a script: apk add somepackages > install.log
2015-06-22 08:33:40 <ncopa> what will the colored output do with the logfile?
2015-06-22 08:33:44 <ncopa> answer: mess it up
2015-06-22 08:34:21 <ncopa> and for problem 3, which is not really a big problem
2015-06-22 08:34:41 <ncopa> but the reasoning is, why slow down things if the message is not read anyway...
2015-06-22 08:36:30 <clandmeter> i guess for 2 apk should run in scripted mode and do not display those messages.
2015-06-22 08:36:43 <clandmeter> and colors should be banned anyways
2015-06-22 08:37:08 <clandmeter> but thats already done with README.alpine is txt only
2015-06-22 08:37:52 <ncopa> we could have an apk-readme trigger:
2015-06-22 08:38:20 <clandmeter> yes, that sounds good.
2015-06-22 08:38:35 <clandmeter> and only exec if we are sure a human is watching :)
2015-06-22 08:38:47 <ncopa> triggers="apk-readme.trigger=/usr/share/doc/*/README.alpine"
2015-06-22 08:39:02 <ncopa> and the script could look something like:
2015-06-22 08:39:40 <ncopa> for i; do echo "### $1"; cat $1; done
2015-06-22 08:40:26 <ncopa> if we want colors for it, then we need some is_tty function
2015-06-22 09:20:54 <kaniini> of course, does anyone actually read maintainer docs?
2015-06-22 09:21:20 <kaniini> i mean, they didn't in debian, so i would suspect the trend continues here for the most part
2015-06-22 09:43:21 <ahills> ncopa: it would be great if apk kept a text log á lá pacman and included post-install notes there, for easy review
2015-06-22 09:44:02 <ncopa> kaniini: thats why i suggest simply remove it :)
2015-06-22 09:44:19 <ahills> Post-install notes are 90% of the time one line, rc-update add X, so just a variable should be sufficient imo
2015-06-22 09:44:42 <ncopa> there was some patches sent to apk-tools for creating log
2015-06-22 09:44:50 <ncopa> i think an apk log would be nice
2015-06-22 09:45:00 <ahills> It would really help when I fuck up my system bad :)
2015-06-22 09:45:01 <ncopa> but i am not sure we want the post-install msg there (if any)
2015-06-22 09:45:50 <ahills> I would like them in a log format somewhere, I often parse logs by date to find out when I did something to resume the task
2015-06-22 09:46:46 <ahills> Most of the rc-update messages are useless but it's quite helpful when the service name differs from the package name
2015-06-22 09:48:33 <ncopa> i wonder if i should do eudev now
2015-06-22 09:48:44 <ncopa> i think we want eudev for v3.3
2015-06-22 09:49:14 <ahills> It's on my to do list after resume from encrypted disk when I get back
2015-06-22 09:49:25 <ahills> I have the package just haven't tested it
2015-06-22 09:49:40 <ncopa> ahills: may i see the APKBUILD?
2015-06-22 09:49:44 <ncopa> i have an APKBUILD too
2015-06-22 09:49:52 <ncopa> i wonder if we should keep the /usr split
2015-06-22 09:49:56 <ahills> Well... Can you wait until tomorrow?
2015-06-22 09:50:00 <ncopa> sure np
2015-06-22 09:50:18 <ncopa> we could say: if you want use udev you may not have /usr in spearate partition
2015-06-22 09:50:23 <ncopa> would make things easier for us
2015-06-22 09:50:25 <ahills> It's also my first day of work at the new job but I'll try to throw it up before I leave for work
2015-06-22 09:50:46 <ncopa> congrats with new job then :)
2015-06-22 09:50:49 <ahills> Hmm, yes, mdev still the default then?
2015-06-22 09:50:52 <ahills> Thanks!
2015-06-22 09:50:54 <ncopa> yes
2015-06-22 09:50:58 <ncopa> mdev would still be the default
2015-06-22 09:51:01 <ahills> OK good
2015-06-22 09:51:14 <ncopa> and in long term, i'd like have an udev replacement
2015-06-22 09:51:16 <ahills> Maybe it's better to put all eudev in /lib
2015-06-22 09:51:19 <ahills> Not /usr
2015-06-22 09:51:31 <ncopa> putting it in /usr would just be convenient
2015-06-22 09:51:35 <ahills> I think I may have done that but no promises, my memory us bad
2015-06-22 09:52:01 <ahills> How would it be more convenient than not-usr?
2015-06-22 09:52:13 <ncopa> thats what upstream does
2015-06-22 09:52:16 <ncopa> i thought
2015-06-22 09:52:16 <ahills> I would rather have it mirror mdev as much as possible I think
2015-06-22 09:52:32 <ahills> Hmmm, well, I will find out tomorrow morning
2015-06-22 09:52:37 <ncopa> but it looks like gentoo does --enable-split-usr
2015-06-22 09:52:43 <ahills> Probably about lunchtime for you
2015-06-22 09:52:49 <ncopa> so it will probably be easier with that
2015-06-22 13:00:46 <clandmeter> ncopa: i updated mirrors package, anyway to update mirros.txt on mirror?
2015-06-22 13:01:37 <ncopa> oh
2015-06-22 13:01:42 <ncopa> looks like its manual
2015-06-22 13:02:13 <ncopa> ok
2015-06-22 13:02:19 <ncopa> i'll make a symlink
2015-06-22 13:02:39 <ncopa> done
2015-06-22 13:03:28 <ncopa> it now use the file from alpine-mirrors @ v3.2
2015-06-22 13:04:14 <ncopa> maybe we should provide a mirrors.json and mirrors.yaml too?
2015-06-22 13:15:53 <tdtrask> ncopa: is the git hook present for creating source packages now that you've reworked git?
2015-06-22 13:16:09 <tdtrask> acf-iptables-0.7.1.tar.xz wasn't created
2015-06-22 13:16:31 <tdtrask> no acf-iptables-0.7.1.tar.xz file was created
2015-06-22 13:16:47 <tdtrask> (sorry for the repeat)
2015-06-22 13:16:58 <tdtrask> (irc client is acting flaky)
2015-06-22 13:19:37 <ncopa> hum
2015-06-22 13:19:43 <ncopa> i thought it worked...
2015-06-22 13:22:49 <tdtrask> maybe the tag didn't push
2015-06-22 13:22:57 <ncopa> tdtrask: there are no tag http://git.alpinelinux.org/cgit/acf/acf-iptables/
2015-06-22 13:23:02 <ncopa> git push --tags
2015-06-22 13:23:21 <tdtrask> yeah, my process was messed up because the repo URL changed
2015-06-22 13:23:40 <tdtrask> 
ACTION uses a script to bump version and push tag

2015-06-22 13:23:59 <tdtrask> now it's created, but a different filename
2015-06-22 13:24:10 <tdtrask> which means the APKBUILD files need to change :(
2015-06-22 13:24:38 <ncopa> thats a bug
2015-06-22 13:25:58 <ncopa> i fixed it manually
2015-06-22 13:27:41 <ncopa> i need to fix that
2015-06-22 13:31:03 <tdtrask> thanks
2015-06-22 13:34:28 <ScrumpyJack> what provides libqt? q5-qtbase-dev?
2015-06-22 13:46:05 <ncopa> if its libqt 4 then its qt
2015-06-22 13:46:16 <ncopa> if its libqt5 the ints qt5-qtbase i think
2015-06-22 13:47:21 <ncopa> i wonder if we should create the dev.a.o/archives in subdirs
2015-06-22 13:47:31 <ncopa> eg dev.a.o/archive/hosted/sircbot
2015-06-22 13:47:44 <ncopa> and dev.a.o/archive/acf/acf-*
2015-06-22 13:47:54 <ncopa> and dev.a.o/archive/user/ncopa/...
2015-06-22 13:48:07 <ncopa> currently they are all just created flat in /archive/
2015-06-22 13:48:25 <ncopa> so if user1 has /user/user1/somerepo
2015-06-22 13:48:37 <ncopa> and user2 has a repo with same name
2015-06-22 13:48:52 <ncopa> then will they overwrite the releases in /archive/somerepo/
2015-06-22 14:01:13 <systmkor> ncopa: I'm compiling gcc on RPi, I glanced through the APKBUILD file and it looked like there was a way of just compiling gcc & g++ for bootstrapping rather than adding in Fortran Ada Java etc., how do i do that?
2015-06-22 14:03:39 <ncopa> looks like you need set LANG_*=false
2015-06-22 14:03:41 <ncopa> so
2015-06-22 14:04:22 <ncopa> LANG_OBJC=false LANG_JAVA=false LANG_GO=false LANG_FORTRNA=false LANG_ADA=false abuild -r
2015-06-22 14:04:39 <tdtrask> ncopa: dev.a.o idea sounds like a better design, but it creates a bunch of work to update existing aports
2015-06-22 14:04:52 <ncopa> tdtrask: thats is the issue
2015-06-22 14:04:53 <tdtrask> wasn't there a comment last week about using cgit rather than dev.a.o?
2015-06-22 14:05:10 <ncopa> tdtrask: there was a comment
2015-06-22 14:05:11 <tdtrask> did they fix the issue with regenerating the zip?
2015-06-22 14:05:16 <ncopa> no
2015-06-22 14:05:17 <ncopa> well
2015-06-22 14:05:32 <ncopa> they have fixed the timestamp in gzip
2015-06-22 14:05:46 <ncopa> so as log we dont update cgit it works
2015-06-22 14:06:04 <ncopa> but last time i relied on that and they updated cgit
2015-06-22 14:06:09 <ncopa> they used a new git version
2015-06-22 14:06:19 <tdtrask> sounds brittle
2015-06-22 14:06:20 <ncopa> that apparently created the tarball differently
2015-06-22 14:06:44 <tdtrask> what's the confidence level they won't break it again moving forward?
2015-06-22 14:06:44 <systmkor> ncopa: thanks :D
2015-06-22 14:06:53 <ncopa> even changing the order of the files makes the tar.gz checksum differ
2015-06-22 14:07:04 <ncopa> so i think the real fix for that in cgit would be cache the tarballs
2015-06-22 14:07:17 <ncopa> basically what we do with git hook now
2015-06-22 14:07:26 <systmkor> is there a way with abuild to recursively build like `abuild -R` but without deleting progress within the current src tree which happens to be gcc atm
2015-06-22 14:08:16 <systmkor> i was thinking -k would do it but, the help suggested it just keeps exisintg packages not object code within a package
2015-06-22 14:09:26 <tdtrask> ncopa: since dev.a.o caches projects that are not in git.a.o, sounds like we need to keep it around anyway
2015-06-22 14:09:37 <tdtrask> so, may as well stick with the git hook
2015-06-22 14:10:03 <tdtrask> is the git hook automatically run for user repos? or manually added to certain repos?
2015-06-22 14:10:07 <ScrumpyJack> while building something, libtool linking fails - like this -  libtool: link: gcc -o a.o b.o c.o etc with the error can't find any.o
2015-06-22 14:10:21 <ScrumpyJack> not sure how to fix that
2015-06-22 14:15:52 <ncopa> tdtrask: the hook is run for every repo
2015-06-22 14:16:04 <ncopa> but that tarball is only generated if the target directory exists
2015-06-22 14:16:41 <ncopa> ScrumpyJack: how do you build the .o files?
2015-06-22 14:16:54 <ncopa> gcc -c file.o file.c
2015-06-22 14:16:58 <ncopa> is the normal way to do it
2015-06-22 14:17:16 <ncopa> and those needs to be built before you link
2015-06-22 14:17:37 <ncopa> so dependencies needs to be done correctly in Makefile
2015-06-22 14:17:52 <ncopa> or you can get that error if you build on multicore
2015-06-22 14:18:10 <ncopa> ScrumpyJack: does it work if you do: make -j1 ?
2015-06-22 14:18:36 <ncopa> if it does, then it is a dependency issue in makefile, causing parallell build fail
2015-06-22 14:21:21 <ncopa> tdtrask: i wonder if we should just do the subdirs properly when generating the tarballs
2015-06-22 14:21:30 <ncopa> hm
2015-06-22 14:21:32 <ncopa> actually
2015-06-22 14:21:35 <tdtrask> 
ACTION is ok with that

2015-06-22 14:21:38 <ncopa> maybe not worth it
2015-06-22 14:21:59 <ncopa> because, imho, the longterm fix is cached tarballs in cgit
2015-06-22 14:22:21 <ncopa> looked at the code it does not look overly difficult to implement
2015-06-22 14:22:31 <ScrumpyJack> ncopa: that sounds about right
2015-06-22 14:23:09 <ncopa> i think we just keep the tarballs in their current location
2015-06-22 14:23:21 <ncopa> if it becomes a problem, we can reconsider
2015-06-22 14:23:25 <ScrumpyJack> none of the .o files are there, so it's linking a bit early i guess
2015-06-22 14:23:35 <ncopa> and for long-term solution i think we want cgit fixed
2015-06-22 14:23:54 <ncopa> ScrumpyJack: try make -j1 then
2015-06-22 14:24:17 <ScrumpyJack> pl
2015-06-22 14:24:23 <ScrumpyJack> what?
2015-06-22 14:24:27 <ScrumpyJack> i meant ok
2015-06-22 14:24:50 <ncopa> :D
2015-06-22 14:27:55 <ScrumpyJack> ncopa: same issue after make -j1 added in APKBUILD file :(
2015-06-22 14:28:24 <ncopa> then it is other problem
2015-06-22 14:32:13 <ScrumpyJack> do you want to take a look?
2015-06-22 14:55:39 <ncopa> maybe tm
2015-06-22 14:56:26 <ncopa> tdtrask: what do you think, should we always create the tarballs? and if target dir is missing, then we create it?
2015-06-22 14:56:33 <ncopa> i think that would be a good idea
2015-06-22 14:56:54 <ncopa> the drawback is that there are no opt-out
2015-06-22 15:19:19 <tdtrask> ncopa: I've forgotten the manual stop before when creating a new repo
2015-06-22 15:19:24 <tdtrask> so, automation might be nice
2015-06-22 15:19:45 <tdtrask> but, not necessary unless you're already making changes
2015-06-23 02:47:28 <mitchty> ok, so i know the -nopie bit in the linker was at issue in the preliminary review from irc a few days ago, but i've compared against the gentoo/fedora ports of ghc and they disable stack smashing protection and disable pie in their hardened configs, so I think its not a huge issue for a preliminary port, so my proposed patch for aports is https://github.com/mitchty/aports/commit/55705ad1e772f4b6519ce8b9e7483d33c3f0a5e0.patch
2015-06-23 02:47:28 <mitchty>  (its not perfect but it compiles itself), and https://github.com/mitchty/alpine-linux-ghc-bootstrap is how I got from a cross compiler to the apk's I've validated, input appreciated
2015-06-23 02:49:51 <mitchty> going to put up the apk's i've rebuilt from themselves up on s3 along with my public signing key and see if my fellow haskellers find issue with the port
2015-06-23 06:13:40 <ncopa> morning
2015-06-23 06:13:46 <ncopa> mitchty: wow, good job
2015-06-23 06:17:57 <kaniini> okay
2015-06-23 06:34:33 <ncopa> hi kaniini
2015-06-23 06:34:50 <ncopa> how is the raid controlleler manager project going?
2015-06-23 06:34:50 <kaniini> hi
2015-06-23 06:35:03 <kaniini> stalled atm, busy work $dayjob
2015-06-23 06:35:43 <ncopa> ok :-/
2015-06-23 06:35:51 <ncopa> those $dayjobs kills the fun :)
2015-06-23 06:51:16 <ncopa> fabled: what do you think about the ghc stuff ^^^
2015-06-23 06:51:26 <fabled> ?
2015-06-23 06:51:31 <fabled> sorry i've lost backlog on that
2015-06-23 06:51:55 <ncopa> i sent to you privately
2015-06-23 06:56:28 <ncopa> hum
2015-06-23 06:56:55 <ncopa> i think maybe i should just move those "build uploaded" to #alpine-commits instead
2015-06-23 06:57:13 <ncopa> it means that the errors goes there too for now
2015-06-23 08:46:29 <ncopa> bah
2015-06-23 08:46:38 <ncopa> pmacct exit thinkgs witih SIGKILL
2015-06-23 09:36:54 <ncopa> rnalrd: fyi. i have a patch for pmacct
2015-06-23 09:37:00 <ncopa> stack size is too msall
2015-06-23 09:37:43 <ncopa> thats why it segfaults #4772
2015-06-23 09:37:51 <ncopa> #4272
2015-06-23 09:52:58 <ncopa> rnalrd: ok? http://sprunge.us/FBhQ
2015-06-23 11:30:39 <ncopa> I am adding a feature to abuild for makeing $pkgname-static subpackages
2015-06-23 11:31:23 <ncopa> this means that normally, the *.a static libs should be shipped with a separate subpackage
2015-06-23 11:31:26 <ncopa> my question is
2015-06-23 11:31:38 <ncopa> currently -dev will get any .a files
2015-06-23 11:31:59 <ncopa> do we want continue to do that, in case there are no -static subpackage=
2015-06-23 11:32:01 <ncopa> ?
2015-06-23 11:32:52 <ncopa> otherwise, if there exists static lib but the -static subpackage is missing, it will no longer go to -dev package
2015-06-23 11:33:13 <ncopa> the alternative is that the $pkgname-static always must be defined before $pkgname-dev
2015-06-23 11:33:34 <ncopa> eg. subpackages="$pkgname-dev $pkgname-static" will not work
2015-06-23 11:34:00 <ncopa> it will be required to be: subpackages="$pkgname-static $pkgname-dev"
2015-06-23 11:34:19 <ncopa> i dont know which is worse
2015-06-23 11:34:23 <ncopa> or best
2015-06-23 11:34:54 <ncopa> i could display a warning if there are a .a file that it should go to -static subpackage
2015-06-23 11:35:36 <ncopa> then we have libs that are purely static, argp-standalone
2015-06-23 11:52:35 <ttilley> ncopa: ping that i added -z now and SSP-by-default to clang. there's a bug open. sorry it took so long to get to. there's also a patch to the build that builds the tests ahead of time and uses paxmark m on them so the llvm test suite can be re-enabled. http://bugs.alpinelinux.org/issues/4374
2015-06-23 11:54:18 <ttilley> i see that there's another bug saying llvm won't build on armhf... i unfortunately don't have a way to do any testing there
2015-06-23 11:58:05 <ttilley> im amazed how nice llvm is to work with despite being c++. i hate c++...
2015-06-23 12:09:52 <vkrishn> update multitail pls
2015-06-23 12:36:09 <vkrishn> I an unable to ./configure --with-th1-docs --with-th1-hooks in testing/fossil
2015-06-23 12:36:37 <vkrishn> compiles ok, but "fossil serve" does not work
2015-06-23 12:36:53 <vkrishn> seems ok on knoppix
2015-06-23 13:11:25 <ScrumpyJack> just got this:
2015-06-23 13:11:26 <ScrumpyJack> alpine:~# /etc/init.d/mariadb setup * Creating a new MySQL database ...
2015-06-23 13:11:27 <ScrumpyJack> /lib/rc/sh/openrc-run.sh: line 354: mysql_install_db: not found           [ !! ]
2015-06-23 13:33:57 <clandmeter> ScrumpyJack: is that stable or edge?
2015-06-23 13:37:49 <alacerda> what is needed to abuild recognize documentation of a package?
2015-06-23 13:37:50 <alacerda> http://hastebin.com/atibequgos.sm
2015-06-23 13:39:38 <BitL0G1c> docs are not in src/pkgname/usr/share/doc || man || info ?
2015-06-23 13:41:05 <BitL0G1c> sometimes I've had to move them in package() or build()
2015-06-23 13:48:49 <ncopa> alacerda: subpackages="$pkgname-doc" only makes sense if you have /usr/share/{man,doc,info} or similar
2015-06-23 13:49:50 <alacerda> I am taking this as an example: http://git.alpinelinux.org/cgit/aports/tree/main/abiword/APKBUILD
2015-06-23 13:50:16 <ScrumpyJack> clandmeter: stable
2015-06-23 13:50:44 <ncopa> its a bug then
2015-06-23 13:52:00 <clandmeter> i think i patched it in edge
2015-06-23 13:52:20 <ScrumpyJack> ok cool, thanks
2015-06-23 13:54:03 <clandmeter> hmm doesnt look like it
2015-06-23 13:57:26 <clandmeter> ScrumpyJack: could be http://git.alpinelinux.org/cgit/aports/commit/main/mariadb?id=106a5eae5aada153c5e806fc2579466f0eca58dd
2015-06-23 14:04:47 <ScrumpyJack> yes, could be
2015-06-23 14:05:01 <ScrumpyJack> "+CONFIGURE_FILE(${CMAKE_CURRENT_SOURCE_DIR}/mysql_install_db.sh"
2015-06-23 14:22:09 <alacerda> ncopa or clandmeter, please: http://sprunge.us/dXgO
2015-06-23 14:42:09 <ScrumpyJack> midori is missing icons. have i forgotten something? http://st.ilet.to/midori.png
2015-06-23 14:43:00 <ScrumpyJack> alacerda: ooo, a man page?
2015-06-23 14:43:18 <alacerda> =]
2015-06-23 14:44:37 <fcolista> alacerda, applied
2015-06-23 14:44:56 <alacerda> fcolista, thx
2015-06-23 14:45:10 <fcolista> np
2015-06-23 15:38:30 <BitL0G1c> essiv https://en.wikipedia.org/wiki/ESSIV#Encrypted_salt-sector_initialization_vector_.28ESSIV.29 does not seem to be working in cryptsetup
2015-06-23 15:43:06 <BitL0G1c> dialog in edge s/be moved to main - in the version in main --prgbox is broken
2015-06-23 15:45:31 <BitL0G1c> (the current version fixed quite a few bugs)
2015-06-23 17:40:23 <BitL0G1c> essiv works
2015-06-24 13:15:03 <mitchty> so figured out the reason -nopie was needed on my ghc port, will look at how to fix it tonight at home
2015-06-24 13:16:17 <mitchty> long story short i may have found a bug on x86_64
2015-06-24 13:34:37 <ScrumpyJack> spice client seems broken for me in edge :(
2015-06-24 13:37:32 <ScrumpyJack> http://sprunge.us/fIfR
2015-06-24 14:19:45 <ScrumpyJack> [B[B[BWarning: untrusted X11 forwarding setup failed: xauth key data not generated
2015-06-24 14:19:48 <ScrumpyJack> Warning: No xauth data; using fake authentication data for X11 forwarding.
2015-06-24 14:28:32 <ScrumpyJack> oops
2015-06-24 15:32:59 <ScrumpyJack> how do i pull the kernel headers for a build?
2015-06-24 15:33:14 <ScrumpyJack> linux-headers?
2015-06-24 15:40:24 <fabled> yes
2015-06-24 15:43:21 <BitL0G1c> ScrumpyJack - try virt-viewer instead of spicy
2015-06-25 08:00:37 <ScrumpyJack> is webkitgtk2-dev missing a dependancy? http://sprunge.us/UTVB
2015-06-25 08:03:45 <barthalion> rather needs a rebuild
2015-06-25 08:24:35 <ncopa> webkitgtk2-dev?
2015-06-25 08:24:44 <ncopa> use webkitgt-dev
2015-06-25 11:04:01 <ScrumpyJack> i used webkitgtk-dev with application seg faults
2015-06-25 11:04:55 <ScrumpyJack> can you spin up a rebuild of webkitgtk2-dev?
2015-06-25 12:04:18 <ncopa> where do I find webkitgtk2-dev?
2015-06-25 12:04:20 <ncopa> edge?
2015-06-25 12:04:22 <ncopa> v3.2?
2015-06-25 12:25:24 <fcolista> anyone have ever used tr-069 ?
2015-06-25 12:51:47 <ScrumpyJack> webkitgtk2 is in edge
2015-06-25 12:53:12 <ScrumpyJack> i think enlightenment is broken in edge too http://sprunge.us/VOMb
2015-06-25 12:53:58 <ScrumpyJack> well, evas
2015-06-26 06:24:14 <ncopa> morning
2015-06-26 06:24:55 <ncopa> clandmeter: turns out that your mpd patch is correct
2015-06-26 06:25:13 <ncopa> #if defined(__GLIBC__) is the correct thing to do
2015-06-26 06:44:22 <clandmeter> ncopa: moring
2015-06-26 06:44:59 <barthalion> mourning
2015-06-26 06:45:41 <ncopa> clandmeter: i'll add a comment in your patch and simplify it
2015-06-26 06:46:00 <clandmeter> ncopa: i could have just removed that whole ifdef crap if its not going upstream?
2015-06-26 06:47:16 <clandmeter> if you are going to update mpd, can you add nfs and maybe also smb support?
2015-06-26 06:53:44 <ncopa> ok
2015-06-26 06:54:03 <ncopa> clandmeter: since you actually use it, do you mind take maintainership?
2015-06-26 06:54:52 <clandmeter> ok
2015-06-26 06:55:03 <ncopa> i can push the change thought
2015-06-26 06:55:06 <ncopa> since i have it done
2015-06-26 06:56:56 <clandmeter> sure just push it. ill add it myself
2015-06-26 06:57:05 <ncopa> i added you already
2015-06-26 06:57:08 <ncopa> thanks!§
2015-06-26 06:57:20 <clandmeter> ah yes is see it
2015-06-26 06:57:32 <clandmeter> did you find your kernel issue?
2015-06-26 07:01:48 <ScrumpyJack> morning
2015-06-26 07:02:32 <ScrumpyJack> mpd? cool
2015-06-26 07:10:32 <ncopa> yes
2015-06-26 07:10:46 <ScrumpyJack> you use it a lot?
2015-06-26 07:12:11 <ScrumpyJack> how do you use it? I does it do something you particularly like?
2015-06-26 07:14:26 <clandmeter> if you are talking to me, i use it as radio/media player in our office. mpd with ympd on an old alix with usb soundcard (or use rpi)
2015-06-26 07:15:41 <ScrumpyJack> so the sound card is on the server? and you control it with clients locally?
2015-06-26 07:16:13 <clandmeter> ympd is a web client
2015-06-26 07:16:35 <clandmeter> i use mpc via cron to automatically switch radio channels
2015-06-26 07:16:56 <clandmeter> radio streams :)
2015-06-26 07:20:23 <clandmeter> I was looking in a sonos alternative previously, but there seems no *easy* way to set it up, although mpd almost has all features buildin.
2015-06-26 07:21:45 <ncopa> cool :)
2015-06-26 07:22:11 <ncopa> there are also mpd client for android i think
2015-06-26 07:22:14 <ncopa> i should set that up for home
2015-06-26 07:22:39 <clandmeter> you want to be maintainer again? :p
2015-06-26 07:22:44 <ncopa> no :)
2015-06-26 07:23:04 <clandmeter> would be nice if mpd could play in sync with other clients
2015-06-26 07:24:47 <barthalion> it doesn't?
2015-06-26 07:25:18 <clandmeter> i dont think so
2015-06-26 07:26:16 <clandmeter> the slidest offset will make 2 speakers sound weird
2015-06-26 07:26:59 <clandmeter> but im not following mpd development, so maybe it has support now?
2015-06-26 07:27:47 <clandmeter> i saw some alterntaive setups which use Pulseaudio
2015-06-26 07:28:03 <clandmeter> 
ACTION runs aways after mentioning PA

2015-06-26 08:13:36 <ncopa> http://build.alpinelinux.org/
2015-06-26 08:15:39 <ScrumpyJack> nothing :(
2015-06-26 08:16:30 <ScrumpyJack> i can see it's A record, but nothing back from an http get
2015-06-26 08:16:37 <ncopa> bah
2015-06-26 08:16:39 <ncopa> stupid me
2015-06-26 08:17:29 <ncopa> i only updated my /etc/hosts file :)
2015-06-26 08:18:52 <ScrumpyJack> ok, now the A record has changed
2015-06-26 08:19:23 <ScrumpyJack> but still nothing :(
2015-06-26 08:19:29 <ncopa> point to .29?
2015-06-26 08:19:32 <ScrumpyJack> yup
2015-06-26 08:19:52 <ScrumpyJack> it's my cache
2015-06-26 08:19:55 <ncopa> does http://distfiles.alpinelinux.org work?
2015-06-26 08:20:22 <ncopa> yes it might need an hour to timout cache
2015-06-26 08:21:31 <ScrumpyJack> so, what am I looking at. the build server? i see sources and logs
2015-06-26 08:24:16 <ScrumpyJack> quite a few sources not saved with any helpful name. probably git releases, and some of those are probably my mistakes :(
2015-06-26 08:41:44 <ncopa> http://build.alpinelinux.org/ is build server status
2015-06-26 08:41:53 <ncopa> eg which build servier is doing what
2015-06-26 08:41:55 <ncopa> realtime
2015-06-26 08:43:03 <barthalion> fancy
2015-06-26 08:43:42 <ncopa> i think next thing is collecting build logs there too
2015-06-26 08:44:00 <ncopa> and publish build errors with exact links to logs
2015-06-26 08:54:53 <Nemykal> getting an issue with syslinux in edge on x86_64. i run `apk fix syslinux` and get: /sbin/update-extlinux: line 93: 3\*10: syntax error: invalid arithmetic operator (error token is "\*10")
2015-06-26 08:55:19 <Nemykal> hmm actually i bet thats because my sh is now bash
2015-06-26 08:55:48 <Nemykal> yeah, it was. nevermind.
2015-06-26 08:55:59 <ncopa> we could try fix it anyway
2015-06-26 08:56:07 <ncopa> making it more portable
2015-06-26 08:56:28 <Nemykal> yeah... i made my sh bash for some other reason, which i can't remember :)
2015-06-26 09:00:02 <ncopa> http://pubs.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#tag_02_06_04
2015-06-26 09:00:25 <Nemykal> also while im here... what package provides gcc plugins in alpine? went to compile a custom kernel and got the error about DISABLE_PAX_PLUGINS
2015-06-26 09:01:08 <ncopa> Nemykal: i think you need some more in makedepends
2015-06-26 09:01:44 <ncopa> http://git.alpinelinux.org/cgit/aports/commit/?id=3a062aad5ee6f761e696d8ed520b1534aff7f97b
2015-06-26 09:02:08 <ncopa> we talked about add a gcc-plugins-dev subpackage which pulls in the needed deps
2015-06-26 09:02:14 <Nemykal> aha
2015-06-26 09:02:17 <Nemykal> Got it, that worked
2015-06-26 09:02:34 <Nemykal> Yeah, I was expecting there to be something with either `gcc` or `plugin` in the pkgname, but it was mpc instead
2015-06-26 09:03:11 <ncopa> kaniini: any objection to http://sprunge.us/PWiE
2015-06-26 09:03:39 <ncopa> Nemykal: yes, it made me lose some hair..
2015-06-26 09:23:07 <clandmeter> ncopa: Nice!
2015-06-26 09:26:44 <clandmeter> ncopa: i dont get the msg's in activity?
2015-06-26 09:29:59 <clandmeter> i see all those pkgs scroll by. I guess they are not build, maybe installed? althought some of them make no sence as deps.
2015-06-26 09:30:05 <ncopa> its the build log
2015-06-26 09:30:23 <ncopa> but the build script is slightly buggy
2015-06-26 09:30:33 <ncopa> it lists some of the packages it should not try build
2015-06-26 09:30:49 <ncopa> i think i'll try clean that up after v3.2.1
2015-06-26 09:49:28 <ScrumpyJack> ncopa: i would have thrown mpd out of the window with a dev like that. plenty of competing software out there. Thanks for still supporting it though
2015-06-26 09:50:23 <ncopa> he probably only had a bad day
2015-06-26 09:50:26 <ncopa> or bad week
2015-06-26 09:50:31 <ncopa> maybe bad year...
2015-06-26 09:50:32 <ncopa> :)
2015-06-26 09:50:33 <clandmeter> bad wife
2015-06-26 09:50:39 <ncopa> bad life :)
2015-06-26 09:50:40 <ScrumpyJack> bad life
2015-06-26 09:50:50 <ScrumpyJack> heh, ninja's by ncopa
2015-06-26 10:33:36 <crow> I was hoping that here will be core created, but seems it is not: [3810220.800510] serve_process[30751]: segfault at 0 ip 00000000cb8154ab sp 00007f3dcfb34a00 error 4 in oscam[cb7d2000+e3000], how to set ulimit to user running this process if user does not have shell?
2015-06-26 11:03:49 <kaniini> ncopa: none
2015-06-26 11:03:55 <kaniini> ncopa: but i've only lightly tested the patch
2015-06-26 11:05:11 <ncopa> kaniini: thanks
2015-06-26 11:05:38 <ncopa> crow: is it an init.d script? add rc_ulimit="-c unlimited" in /etc/conf.d/oscam or similar
2015-06-26 11:14:10 <ScrumpyJack> hmm, can omxplayer be windowed in tmux?
2015-06-26 11:17:31 <ScrumpyJack> or anything for that matter
2015-06-26 13:26:07 <leo-unglaub> hey guys
2015-06-26 13:27:27 <ncopa> hi
2015-06-26 13:33:53 <leo-unglaub> a quick question about how you build your packages, i am using a virtual machine for it currently, but thats slow as hell. i want to do it on my main maschine but i dont want all dev dependencies bloading up my system. has someone of you a ready working build in chroot script?
2015-06-26 13:35:01 <crow> ncopa yes it is, ill try
2015-06-26 13:36:49 <ncopa> leo-unglaub: i use lxc
2015-06-26 13:36:51 <ncopa> containers
2015-06-26 13:36:58 <crow> ncopa like this http://sprunge.us/NiMP ?
2015-06-26 13:37:20 <ncopa> crow: yes, like that
2015-06-26 13:37:43 <ncopa> leo-unglaub: what do you use as your main machine?
2015-06-26 13:38:00 <leo-unglaub> ncopa: currently virtualbox ... but that was a huge mistake ...
2015-06-26 13:38:12 <leo-unglaub> just updates firefox aurora and it took me 3 hours to build
2015-06-26 13:38:42 <leo-unglaub> ncopa: ah, you dont meat for building?
2015-06-26 13:38:56 <leo-unglaub> my main maschine is alpine on and intel i7 with SSD raid 0 for extra speed
2015-06-26 13:39:18 <crow> ncopa and the cat /proc/sys/kernel/core_pattern returns me "core" which mean what? the core dump will be created there where executable is?
2015-06-26 13:41:58 <ncopa> leo-unglaub: i was thinking main machine distro
2015-06-26 13:42:09 <ncopa> what distro you run as virtualbox host
2015-06-26 13:42:28 <leo-unglaub> ncopa: i use alpine for everything. Desktop, laptop, servers
2015-06-26 13:42:38 <ncopa> oh cool :)
2015-06-26 13:42:44 <leo-unglaub> i only boot windows when i play computer games
2015-06-26 13:42:53 <leo-unglaub> or to run virtualbox if i want to try some stuff
2015-06-26 13:43:00 <leo-unglaub> but other than that, alpine only
2015-06-26 13:43:21 <ncopa> I use lxc heavily for development
2015-06-26 13:43:56 <ncopa> i use alpine as main desktop, but do the aports development in a lxc container
2015-06-26 13:44:29 <leo-unglaub> i see, i will read up on lxc, because i am not very familliar with this kind of thing
2015-06-26 13:44:34 <leo-unglaub> but thanks for the tip!
2015-06-26 13:44:49 <ncopa> apk add lxc lxc-templates && lxc-create -t alpine -n edgebox -- --release edge
2015-06-26 13:45:17 <ncopa> lxc-create -t alpine -n dev3.2 -- release v3.2
2015-06-26 13:45:37 <ncopa> it is like chroot on steroids
2015-06-26 13:46:05 <ncopa> like something between virtualbox and chroot
2015-06-26 13:46:27 <leo-unglaub> ah, so i can put that on a ramdisk? and have nice compile speed?
2015-06-26 13:46:58 <ncopa> if you havs ssd raid0 you already have nice compile speed :)
2015-06-26 13:47:15 <ncopa> i have been thinking of tmpfs containers
2015-06-26 13:47:23 <leo-unglaub> ncopa: not really, because i use luks on it. that acrtually slows it down 5%
2015-06-26 13:47:32 <ncopa> but i have not implemented it
2015-06-26 13:47:35 <leo-unglaub> so memory is faster
2015-06-26 13:48:14 <ncopa> you can probably mount /var/lib/lxc as tmpfs
2015-06-26 13:48:22 <ncopa> but then you'll lose all your containers on reboot
2015-06-26 13:49:05 <crow> ncopa would this be enough core creating dump? http://sprunge.us/aQfY
2015-06-26 13:49:23 <ncopa> i have also been thinking of adding support for changing workdir for abuild
2015-06-26 13:49:33 <ncopa> so you could point your abuild workdir to a tmpfs dir
2015-06-26 13:49:44 <ncopa> shouldnt be too hard to implement
2015-06-26 13:49:57 <ncopa> that way you could have all your buliding in tmpfs
2015-06-26 13:50:17 <leo-unglaub> that would be great!
2015-06-26 13:50:37 <ncopa> crow: i think so
2015-06-26 13:51:42 <leo-unglaub> ncopa: so basically  _builddir would be absolute?
2015-06-26 13:51:52 <ncopa> no
2015-06-26 13:52:01 <ncopa> well
2015-06-26 13:52:02 <ncopa> yes
2015-06-26 13:52:06 <ncopa> sorry yes
2015-06-26 13:52:18 <ncopa> i think srcdir is already?
2015-06-26 13:52:57 <leo-unglaub> i have to admin, i would have to look it up. i dont know that from the top of my mind right now
2015-06-26 13:53:13 <ncopa> i think it is
2015-06-26 13:53:18 <ncopa> anyway
2015-06-26 13:53:24 <ncopa> i can look at that when i get a chance
2015-06-26 13:53:29 <ncopa> after v3.2.1 is out
2015-06-26 13:54:03 <leo-unglaub> ncopa: another thing that would be very usefull would be auto applying .patch files in the package directory
2015-06-26 13:54:15 <leo-unglaub> i always have to look up the code on how to use patch correctly *g*
2015-06-26 13:54:20 <ncopa> yeah
2015-06-26 13:54:26 <ncopa> we have lots of copy/paste of that
2015-06-26 13:54:31 <ncopa> in N APKBUILDs
2015-06-26 13:54:56 <ncopa> thats a bigger change though
2015-06-26 13:55:06 <ncopa> i have been thinking of replaceing APKBUILD format
2015-06-26 13:55:10 <ncopa> make a new format
2015-06-26 13:55:28 <ncopa> to address things like that
2015-06-26 13:55:42 <ncopa> but its a big change
2015-06-26 13:55:53 <ncopa> and i am not sure if we want replace it or improve it
2015-06-26 13:56:04 <leo-unglaub> ncopa: if you think of changing the APKBUILD format, may i make a suggestion?
2015-06-26 13:57:06 <leo-unglaub> if you ever change the format, then please but the checksums in a seperate file *g*
2015-06-26 13:57:23 <leo-unglaub> it would make looking thru git commits so much easier *g*
2015-06-26 13:57:31 <ncopa> hum
2015-06-26 13:57:40 <kaniini> would you like us to put in an alpine/ dir
2015-06-26 13:57:40 <kaniini> with alpine/rules
2015-06-26 13:57:43 <kaniini> and etc.
2015-06-26 13:57:48 <ncopa> but it would double the number of files in the aports tree
2015-06-26 13:57:51 <kaniini> 
ACTION ducks.

2015-06-26 13:58:09 <ncopa> many small files -> slow processing the tree
2015-06-26 13:58:50 <ncopa> but i see your point though
2015-06-26 13:59:02 <leo-unglaub> i think that change would be worth it. because currently if you look thru changes on a buildfile you always get changes on the checksums as well, even if you only want to check chen a compil flag changes
2015-06-26 14:00:01 <ncopa> hum
2015-06-26 14:00:19 <ncopa> well i think its a valid reason to move the checksums out
2015-06-26 14:00:27 <ncopa> makes git logs nicer
2015-06-26 14:01:01 <ncopa> i would like move the install scripts in to the master file
2015-06-26 14:01:33 <fabled> ncopa, just removing sha1 + sha256 in favor of sha512 should be ok enough
2015-06-26 14:01:41 <fabled> i'd rather keep them in same file
2015-06-26 14:01:54 <fabled> but having the checksum three times is a bit too much ;)
2015-06-26 14:02:00 <ncopa> :)
2015-06-26 14:02:09 <fabled> and remove checksums if there's gpg/pgp key
2015-06-26 14:02:15 <fabled> use gpg/pgp where available
2015-06-26 14:02:41 <ncopa> hose are good ideas
2015-06-26 14:02:45 <ncopa> those*
2015-06-26 14:03:06 <ncopa> i wonder if we can remove sha1 and sha256
2015-06-26 14:03:12 <ncopa> i suppose we can by now
2015-06-26 14:03:18 <leo-unglaub> yes, you can
2015-06-26 14:03:35 <leo-unglaub> sha512 makes them obsolet
2015-06-26 14:03:55 <ncopa> i was thinkgin the overlap from only support sha1
2015-06-26 14:04:02 <ncopa> but i think thats no problem
2015-06-26 14:04:16 <fabled> it's only abuild requiring it
2015-06-26 14:04:22 <ncopa> yeah
2015-06-26 14:04:30 <ncopa> and all supported branches has sha512
2015-06-26 14:04:42 <ncopa> there are no reason to keep sha1 now
2015-06-26 14:04:52 <fabled> yeah
2015-06-26 14:05:02 <leo-unglaub> if they where seperate files you could now only do a "find . -name *.sha1file -delete" ;)
2015-06-26 14:06:04 <ncopa> i'd not have one file per hash algo anyway
2015-06-26 14:06:24 <ncopa> it would have been a single file with all supported hashalgos
2015-06-26 14:07:10 <leo-unglaub> i see, so not the sha512sum format?
2015-06-26 14:07:30 <ncopa> a file with it ebedded or similar
2015-06-26 14:07:48 <ncopa> [sha512sums]
2015-06-26 14:07:49 <ncopa> ....
2015-06-26 14:07:54 <ncopa> [sha256sums]
2015-06-26 14:07:55 <ncopa> ....
2015-06-26 14:08:09 <ncopa> in same file
2015-06-26 14:08:21 <ncopa> or yaml
2015-06-26 14:08:25 <ncopa> anyways
2015-06-26 14:09:02 <ncopa> simply dropping sha1 and sha256 should be simple
2015-06-26 14:09:28 <ncopa> replacing the apkbuild format is a bigger proejct
2015-06-26 14:09:54 <fabled> the problem is it's easier to audit APKBUILD changes if checksum is same file
2015-06-26 14:09:59 <fabled> otherwise you update one file but not the other
2015-06-26 14:10:55 <ncopa> the new format ideas has been:
2015-06-26 14:10:57 <ncopa> [abuild]
2015-06-26 14:11:02 <ncopa> pkgname=...
2015-06-26 14:11:05 <ncopa> metadata...
2015-06-26 14:11:09 <ncopa> [patches]
2015-06-26 14:11:16 <ncopa> list of patches...
2015-06-26 14:11:20 <ncopa> [build]
2015-06-26 14:11:28 <ncopa> inline shell for building
2015-06-26 14:11:31 <ncopa> ...
2015-06-26 14:11:44 <ncopa> [checksums sha512]
2015-06-26 14:11:54 <ncopa> inline checksums...
2015-06-26 14:12:01 <ncopa> something like that
2015-06-26 14:12:10 <ncopa> i'd also like to inline the install scripts
2015-06-26 14:12:11 <leo-unglaub> so basically a mix of .ini and shell script?
2015-06-26 14:12:15 <ncopa> yeah
2015-06-26 14:12:17 <fabled> ncopa, don't use []
2015-06-26 14:12:19 <fabled> i'd rather go yaml
2015-06-26 14:12:27 <ncopa> thats what i have been thinking
2015-06-26 14:12:28 <ncopa> ok
2015-06-26 14:12:32 <ncopa> :)
2015-06-26 14:12:45 <ncopa> fabled: why not [section] ?
2015-06-26 14:12:57 <fabled> you'll end up bumping to the limits
2015-06-26 14:13:14 <fabled> yaml is more versatile, and more standardized
2015-06-26 14:13:18 <fabled> and we have better libs for it in lua
2015-06-26 14:13:28 <ncopa> hm
2015-06-26 14:13:38 <fabled> you get lists, containers, maps etc. for free
2015-06-26 14:13:39 <ncopa> i was thinking of mixed content
2015-06-26 14:13:42 <fabled> we'll end up using that
2015-06-26 14:13:49 <fabled> there's mixed content users for yaml too
2015-06-26 14:14:07 <ncopa> how do we do shell code in yaml?
2015-06-26 14:14:28 <ncopa> also been thinking of implementing some variant of redhat's .spec format
2015-06-26 14:15:27 <leo-unglaub> hehe, i guess package formats are like raindrops ... if you ask 10 people you get 10 different ideas *g*
2015-06-26 14:15:35 <ncopa> yeah
2015-06-26 14:15:42 <ncopa> it is a non-trivial thing
2015-06-26 14:15:47 <ncopa> but generally
2015-06-26 14:15:52 <ncopa> when we started
2015-06-26 14:16:04 <ncopa> one of the goals was to make it simple to understand
2015-06-26 14:16:07 <ncopa> clear what is going on
2015-06-26 14:16:09 <fabled> for mixing yaml and embedding lua snippets: see https://github.com/gvvaughan/specl/
2015-06-26 14:16:45 <ncopa> now we probably want sacrifice some of the simplicity for simplifying maintenance
2015-06-26 14:16:57 <leo-unglaub> yeah, simple and clean is important ... i hate .deb formats where you need to study for y year just to understand it ...
2015-06-26 14:17:35 <ncopa> finding the balance between simple and easy to maintain is hard
2015-06-26 14:17:44 <ncopa> i didnt want end up as gentoo
2015-06-26 14:18:28 <leo-unglaub> *g*
2015-06-26 14:21:27 <ncopa> we could use lua too
2015-06-26 14:21:41 <ncopa> with a helper function for shell
2015-06-26 14:21:46 <ncopa> sh([[
2015-06-26 14:21:51 <ncopa>  mkdir -p ...
2015-06-26 14:21:54 <ncopa> ]])
2015-06-26 14:22:09 <ncopa> it would set up the environment vars
2015-06-26 14:22:11 <ncopa> hm
2015-06-26 14:32:44 <ncopa> leo-unglaub: did you send patch?
2015-06-26 14:33:19 <leo-unglaub> ncopa: not yet, i am still in the office
2015-06-26 14:33:30 <ncopa> ok, thats why i didnt find it :)
2015-06-26 14:33:51 <leo-unglaub> when my boss is around i always have to pretend to work on something *g*
2015-06-26 14:33:56 <ncopa> :D
2015-06-26 14:34:11 <leo-unglaub> and h knows by now that working in a "black window" aka terminal is mostly not work related *g*
2015-06-26 14:34:24 <ncopa> i use the "compiling" argument
2015-06-26 14:34:38 <ncopa> https://xkcd.com/303/
2015-06-26 14:34:45 <leo-unglaub> ncopa: i tryed that ... but my company builds web api's *g*
2015-06-26 14:34:57 <ncopa> lol
2015-06-26 14:34:58 <leo-unglaub> someone told him that php is interpreted *g*
2015-06-26 14:35:31 <leo-unglaub> beleave me, 3 years ago a simply find / -name '*' was enought to fool him *g*
2015-06-26 14:38:05 <leo-unglaub> brb meeting. see you guys later
2015-06-26 14:39:25 <ncopa> have a nice weekend
2015-06-26 14:43:46 <ncopa> hum patchwork does not pick up patches in stable tree
2015-06-26 15:01:09 <crow> ncopa again new segfault but no core created
2015-06-26 15:02:04 <ncopa> what does dmesg say?
2015-06-26 15:02:12 <ncopa> is there some easy way for me to reproduce it?
2015-06-26 15:03:12 <crow> [3830165.693872] serve_process[20965]: segfault at 0 ip 00000000e28624ab sp 00007f8857fc5d60 error 4 in oscam[e281f000+e3000]
2015-06-26 15:03:22 <crow> well it was start from oscam, but it is not always there
2015-06-26 15:04:15 <crow> i dont know what to do else to have core dump created then your suggestion http://sprunge.us/NiMP and http://sprunge.us/aQfY
2015-06-26 15:04:16 <ncopa> i need to go
2015-06-26 15:04:57 <ncopa> have a nice weekend
2015-06-26 15:05:14 <ncopa> crow: if you have simple steps for me to reproduce it, then please file a bug on bugs.a.o
2015-06-26 15:05:19 <ncopa> and i'll have a look at it
2015-06-26 15:05:47 <crow> have a nice weekend to
2015-06-26 15:05:50 <crow> no not realy.
2015-06-26 15:05:55 <crow> it hanppens
2015-06-26 15:31:13 <stateless>  2. - Πέμπτη " στα Παλιατζίδικα ", ώρα έναρξης 22:30.
2015-06-26 15:31:16 <stateless> Οδός : Νορμανού 3, Μοναστηράκι.
2015-06-26 15:31:19 <stateless> τηλ. κρατήσεων : 697 85 041 93 2. - Πέμπτη " στα Παλιατζίδικα ", ώρα έναρξης 22:30.
2015-06-26 15:31:22 <stateless> Οδός : Νορμανού 3, Μοναστηράκι.
2015-06-26 15:31:22 <stateless> erggh
2015-06-26 15:31:26 <stateless> fail
2015-06-26 15:51:53 <mitchty> ok so now i know why I had to use -nopie to get ghc ported, https://ghc.haskell.org/trac/ghc/ticket/9007 long story short ghc needs to detect when to emit PIC assembly
2015-06-26 18:12:14 <systmkor> mitchty: I need to push the go lang devs to support PIE
2015-06-26 18:13:01 <systmkor> I tried about a year ago but they just said "it's pointless"
2015-06-26 18:13:04 <systmkor> 
ACTION sigh

2015-06-26 18:14:36 <systmkor> hopefully this article can change their minds http://blog.stalkr.net/2015/04/golang-data-races-to-break-memory-safety.html
2015-06-26 18:29:18 <mitchty> systmkor interesting, hadn't heard about that about golang and races, it seems a bit difficult to exploit however
2015-06-26 18:31:43 <systmkor> mitchty: I agree it isn't easy with all the things go lang does do to protect itself
2015-06-26 18:32:03 <systmkor> mitchty: however to me if your code can't be compiled to PIC or PIE your code isn't good
2015-06-26 18:32:20 <systmkor> mitchty: and I mean that more in a your code isn't mature sense
2015-06-26 18:34:50 <systmkor> I grant though I don't have a strong technical argument for that, at the moment
2015-06-26 18:35:40 <mitchty> systmkor oh agree, and the bug I hit is interesting in that its more i need at least a toggle for: just use pic/pie all the time if you could so it is a lot less effort than I think golang will have
2015-06-26 18:36:28 <systmkor> m'kay
2015-06-26 18:36:35 <mitchty> i'm trying to think of c i've written as userspace that would require nopie though
2015-06-26 18:36:41 <systmkor> mitchty: btw thanks for getting haskell on Alpine :D
2015-06-26 18:36:48 <mitchty> i'm drawing a complete blank though tbh
2015-06-26 18:37:35 <mitchty> systmkor no problem, when I saw someone got the gentoo bit working with musl i decided enough was enough >.<
2015-06-26 18:38:42 <mitchty> but i have a few patches I found in the fedora port and another that just got added recently that helps ghc understand -fPIC in its linking stage
2015-06-26 18:38:49 <systmkor> I really need to spend time on getting cross compiling easier (at least for myself) for alpine
2015-06-26 18:39:15 <mitchty> the gcc cross compiler with musl libc was a godsend
2015-06-26 18:39:58 <systmkor> well for me I'm trying to learn cross compiling for arm devices from Alpine
2015-06-26 18:40:10 <mitchty> https://bitbucket.org/GregorR/musl-cross
2015-06-26 18:40:35 <mitchty> i've been using it for some of my c things, its rather nice in general
2015-06-26 18:40:53 <systmkor> noice
2015-06-26 18:41:04 <mitchty> and yep arm ghc was next up on my plate, though i'm not in a rush
2015-06-26 18:41:19 <systmkor> yah as soon as I can wrap my brain around that, I want to make a guide on cross compiling/bootstrapping Alpine Linux to another architecture from Alpine
2015-06-26 18:42:01 <systmkor> then qemu (emulated native) compile
2015-06-26 18:42:10 <systmkor> and finally native compiling after bootstrap
2015-06-26 18:42:42 <mitchty> yep, i've not had to fire up qemu yet to try out arm stuff but should be fun to learn
2015-06-26 18:43:09 <mitchty> i've a radxa rock quad core arm box running debian wheezy that is nice to use
2015-06-26 18:43:21 <mitchty> not enough memory to link some programs though
2015-06-26 18:43:30 <mitchty> even with 2g of memory
2015-06-26 18:44:45 <systmkor> mitchty: I almost was successful compiling gcc natively on RPi B+ but I ran out of disk space, I'm going to try again this weekend
2015-06-26 18:45:40 <mitchty> i just use nfs, it tends to be fast enough actually
2015-06-26 18:46:25 <mitchty> well with a wired connection, with wireless and nfsv3 i'll withhold judgement
2015-06-26 18:46:28 <systmkor> mitchty: noted and I'll let you know how it goes
2015-06-26 18:46:57 <systmkor> goal is to compile gcc natively twice, from I think it's more generic armhf to native, then native to native
2015-06-26 18:47:21 <mitchty> that has to take forever
2015-06-26 18:47:34 <systmkor> it's actually not that bad, i think it's done over night
2015-06-26 18:47:47 <systmkor> I have spare RPis that I don't need to use at the moment
2015-06-26 18:48:13 <systmkor> well overnight for each compile of gcc
2015-06-26 18:48:31 <systmkor> it ends up taking like over 2 gigs of disk space to compile
2015-06-26 18:49:00 <systmkor> but I don't know how to do the bootstrap abuild of gcc, so it doesn't compile gcc with Ada, Go, Java, Fortran, and Obj-C
2015-06-26 18:49:50 <mitchty> isn't it just --enable-languages=c,c++?
2015-06-26 18:50:09 <systmkor> and since Alpine recent abuild is now building for GCC 5.1, it has OpenMP built in, which possible to create a cluster like distc++
2015-06-26 18:50:24 <systmkor> mitchty: idk, there isn't really any documentation
2015-06-26 18:50:41 <systmkor> I looked at the APKBUILD but at a glance I was worried I would screw something up
2015-06-26 18:51:11 <mitchty> i've built gcc for solaris a few times, i used that often to make it compile quicker
2015-06-26 18:51:28 <mitchty> i'd love to learn ada (really) but i don't need it normally
2015-06-26 18:51:49 <leo-unglaub> ncopa: did you get the patches? i am not quit sure if sending them worked
2015-06-26 18:51:52 <mitchty> didn't know about 5.1 and openmp, still waiting for clang myself to get proper support
2015-06-26 18:52:09 <systmkor> well I need to make sure it works with APKBUILD file since there are so many patches and other tweaks to get it to work for Alpine's build system
2015-06-26 18:52:44 <systmkor> mitchty: yah I wouldn't want to use distc++ or OpenMP until there is a solid twice compiled gcc on an a specific arm board
2015-06-26 18:55:07 <mitchty> systmkor i just want an arm board with 4-8g of ram now to be honest, they take almost no power even at full tilt so I can just wait while it does stuff, distcc could help a lot of the  slowness
2015-06-26 18:59:31 <systmkor> mitchty: worse case 6gigs of swap space :P
2015-06-26 19:00:09 <systmkor> mitchty: well with pre ARMv8 cpus I don't think it can do more than 4 gigs
2015-06-26 19:00:12 <systmkor> since it's 32-bit
2015-06-26 19:00:52 <systmkor> well I forget what the extent of extended memory on 32-bit computers is but yah
2015-06-26 19:03:51 <mitchty> well looks like i can get a rock2 now, gets me to 4g of ram at least
2015-06-26 19:03:59 <mitchty> not cheaply though
2015-06-26 19:04:00 <mitchty> http://www.effiall.com/Rock_2_full_development_p/rr2.htm
2015-06-26 19:06:42 <systmkor> mitchty: I want that so bad
2015-06-26 19:06:43 <systmkor> :P
2015-06-27 02:41:53 <kaniini> mitchty: hi
2015-06-27 02:42:19 <kaniini> mitchty: so i was thinking we could use your docker stuff for initial bringup and then have an APKBUILD which installs the bootstrap ghc
2015-06-27 02:42:48 <kaniini> mitchty: and then have a real ghc which depends on ghc-bootstrap
2015-06-27 03:18:50 <mitchty> kaniini sure, so the way i set up the port was to be honest a bit of a hack, mostly due to me trying to find out why the pie hack was needed so if i get you right you want something like
2015-06-27 03:19:59 <mitchty> bootstrap build on non alpine -> transfer that over to alpine, set it up to compile ghc (need alex/happy to do so), then compile ghc-bootstrap which will likely do the same and use that to build the "proper" ghc
2015-06-27 03:21:59 <kaniini> mitchty: no an APKBUILD which installs the bootstrap package and then builds real ghc with it
2015-06-27 03:24:03 <mitchty> kaniini so straight from bootstrap cross compiler to build ghc itself?
2015-06-27 03:25:07 <kaniini> yeah
2015-06-27 03:25:36 <mitchty> should be possible, that cross compilers slow as molasses though
2015-06-27 03:26:49 <mitchty> but i have a few questions on some packaging details unique to ghc in some regards
2015-06-27 03:27:36 <mitchty> so besides needing to self host, which is annoying but ok, to build itself it needs two packages alex/happy, think lex/bison ish in function
2015-06-27 03:28:51 <mitchty> I dealt with this in a bit of a hacky way by building cabal, which is the gem/pip equivalent for haskell, then using that to install both, though technically its possible to not have to do that, I just didn't want to create 10 different apkbuild packages to get to that point
2015-06-27 03:29:44 <mitchty> so for this bootstrap I presume we would be ok with doing similar, but probably would want to do more what arch linux has in the aur I presume
2015-06-27 03:33:58 <mitchty> i setup the apkbuilds here to depend on themselves to make sure everything was ok, once you have your native compiler the bootstrap bit becomes mostly unnecessary
2015-06-27 03:34:03 <mitchty> err https://github.com/mitchty/alpine-linux-ghc-bootstrap/tree/master/stage3 rather
2015-06-27 10:57:16 <msaq> I heard developers were thinking about giving up on APKBUILD's simplicity and using complex stuff like yaml, lua and even .spec to favor "maintainability". Is this true?
2015-06-27 10:58:24 <msaq> I think this is a radical change and goes off with the siḿplicity ideal that I thought Alpine push forward (one of the last distributions to do so)
2015-06-27 10:58:53 <msaq> What's wrong with the current APKBUILD? Is there no way to "fix" it without adding such complexity?
2015-06-27 11:04:37 <barthalion> spec?
2015-06-27 11:04:39 <barthalion> wat?
2015-06-27 11:04:54 <barthalion> we are merely discussing what is wrong with APKBUILDs now and how could we fix it
2015-06-27 11:05:10 <barthalion> I have no idea where you read about considering spec
2015-06-27 11:05:30 <msaq>  14:14 <ncopa> also been thinking of implementing some variant of redhat's .spec format
2015-06-27 11:05:43 <barthalion> implementing variant != using spec
2015-06-27 11:05:47 <barthalion> anyway, we're currently in the middle of nowhere with that because everything has disadvantages
2015-06-27 11:05:55 <msaq> I see...
2015-06-27 11:06:19 <msaq> Well, I just hope Alpine continues to enforce simplicity over anything else
2015-06-27 11:06:33 <msaq> I guess that's still the main goal?
2015-06-27 11:06:38 <barthalion> yes
2015-06-27 11:06:45 <barthalion> this is why nothing has been decided so far
2015-06-27 11:06:50 <msaq> Nice
2015-06-27 11:06:52 <msaq> :)
2015-06-27 11:07:26 <barthalion> we should probably use some templating system on top of APKBUILDs, but I have no idea what ncopa thinks about it
2015-06-27 11:07:48 <barthalion> and because since 10 months or so I only complain here and do not code, I have no vote privilege :P
2015-06-27 11:08:45 <barthalion> but as I said, it's just a discussion
2015-06-27 11:09:14 <msaq> fabled talked about yaml, a less complex alternative would be json
2015-06-27 11:09:36 <msaq> or even good and old plain-text
2015-06-27 11:09:43 <msaq> what's wrong with that?
2015-06-27 11:10:19 <barthalion> well, anything will be good as long as it's easy to implement, convenient to write in and allows deduplication
2015-06-27 11:10:38 <barthalion> although I guess we are looking in somehow bad direction
2015-06-27 11:10:53 <barthalion> Gentoo uses shell script after all, they are just good abstracted from actual shell
2015-06-27 11:11:07 <Drakevr> ew, json is hideous
2015-06-27 11:11:14 <barthalion> yep, yaml > json
2015-06-27 11:11:32 <Drakevr> yaml does have its issues but would take it anyday over json
2015-06-27 11:12:18 <msaq> can we agree plain text is better than both?
2015-06-27 11:13:35 <msaq> Drakevr, barthalion: json is generally faster though http://stackoverflow.com/questions/2451732/how-is-it-that-json-serialization-is-so-much-faster-than-yaml-serialization-in-p
2015-06-27 11:13:50 <msaq> yaml is more complex
2015-06-27 11:13:59 <barthalion> I wouldn't care much about it
2015-06-27 11:15:24 <Drakevr> appart from the question seeming python parser specific
2015-06-27 11:15:36 <Drakevr> unless the overhead is SO MUCH it would hardly matter
2015-06-27 11:15:59 <msaq> Drakevr: http://www.yaml.org/spec/1.2/spec.html#id2759572
2015-06-27 11:16:12 <msaq> "JSON’s foremost design goal is simplicity and universality. Thus, JSON is trivial to generate and parse, at the cost of reduced human readability."
2015-06-27 11:16:23 <msaq> "In contrast, YAML’s foremost design goals are human readability and support for serializing arbitrary native data structures. Thus, YAML allows for extremely readable files, but is more complex to generate and parse"
2015-06-27 11:17:08 <msaq> The question is: does better human readability justifies the added complexity? Most people would say 'yes'. That's why most distributions are utterly complex.
2015-06-27 11:17:27 <msaq> But I thought Alpine people would be the ones to stand up and say 'no'. Better keep things simple.
2015-06-27 11:17:32 <Drakevr> complexity on the parser, not the format
2015-06-27 11:18:01 <Drakevr> i haven't written code for core alpine nor have read any so i can't be sure how complex things are currently
2015-06-27 11:18:23 <barthalion> the main goal is deduplication of some code we have now
2015-06-27 11:18:35 <barthalion> another one is increased readability, sorry
2015-06-27 11:18:51 <barthalion> json always looked like shit to me :(
2015-06-27 11:19:16 <barthalion> we want more people to contribute
2015-06-27 11:20:18 <barthalion> currently it's achieved due to similarity to PKGBUILDs
2015-06-27 12:32:35 <ahills> APKBUILD format is extremely easy to work with if you are building a package, except for some complex exceptions
2015-06-27 12:33:27 <ahills> I think that complexity could probably be handled if it was changed from the var= style to the OpenRC command style
2015-06-27 12:33:52 <ahills> For e.g. complex subpackage dependencies
2015-06-27 16:46:03 <ahills> ncopa, clandmeter, kaniini, et al: I have hibernation/resume working with grsec and FDE (via suspend-utils), but it requires modification of the initramfs's /init script... are there any plans to add hooks, or should I have a mkinitramfs -i <new script> available in suspend-utils?
2015-06-27 16:46:26 <ahills> I'm mostly excited that s2both is working
2015-06-27 16:54:07 <bCQc0P3DIqbh> good job.
2015-06-27 16:59:20 <ahills> it wouldn't have taken this long had I not gone on vacation, silly me
2015-06-27 17:04:57 <bCQc0P3DIqbh> what's the change to init?
2015-06-27 17:07:21 <ahills> calling /usr/lib/suspend/resume with the options from the kernel command line
2015-06-27 17:07:28 <ahills> (or calling it at all)
2015-06-27 17:26:14 <bCQc0P3DIqbh> hmm, interesting, thanks for the info.
2015-06-27 17:41:26 <barthalion> that feeling when one of two last patches on patchwork is assigned to me…
2015-06-27 19:35:08 <barthalion> clandmeter: in which repo is our fancy package searcher held?
2015-06-27 19:35:14 <barthalion> (git repo, that is)
2015-06-27 19:39:31 <k0r10n> hi there, there is chromium APKBUILD's - > https://github.com/k0r10n/aports
2015-06-27 19:39:53 <k0r10n> currently 37 version is working without sandboxing and building successfully on 3.2 Alpine
2015-06-27 19:40:00 <k0r10n> 43 is not ready yet
2015-06-27 19:40:48 <msaq> NICE
2015-06-27 19:41:02 <k0r10n> if somebody want to help - you're welcome !
2015-06-27 19:48:10 <bCQc0P3DIqbh> sandboxing needs code patches to adapt to musl
2015-06-27 20:10:06 <leo-unglaub> ncopa: ping
2015-06-28 11:08:27 <leo-unglaub> good morning :)
2015-06-28 15:22:47 <clandmeter> barthalion: you mean pkgs.a.o?
2015-06-28 15:27:53 <barthalion> yeah
2015-06-28 15:34:01 <clandmeter> it consists of 2 partss
2015-06-28 15:34:41 <clandmeter> http://git.alpinelinux.org/cgit/user/clandmeter/repo-scripts to generate sqlite
2015-06-28 15:35:09 <clandmeter> http://git.alpinelinux.org/cgit/user/clandmeter/aports-turbo the web part
2015-06-28 15:35:27 <barthalion> thx
2015-06-28 15:35:30 <clandmeter> and its far from perfect, but it kind of works
2015-06-28 15:36:22 <clandmeter> barthalion: i think my github is more recent https://github.com/clandmeter/aports-turbo
2015-06-28 17:37:23 <pluesch> http://bugs.musicpd.org/view.php?id=4387
2015-06-28 17:37:49 <pluesch> srsly... why?
2015-06-28 17:38:26 <barthalion> lol
2015-06-28 17:39:05 <pluesch> I thought mpd was cool....
2015-06-28 19:58:51 <vkris> found a minor issue, when just chmod`ing a file say example in /etc/local.d it does not regiter in lbu
2015-06-28 19:59:12 <vkris> register*
2015-06-28 19:59:35 <vkris> 2. why does `apk add iftop` pull atop ;)
2015-06-28 20:56:38 <barthalion> it does not
2015-06-28 20:56:42 <barthalion> http://pkgs.alpinelinux.org/package/main/x86/iftop
2015-06-28 21:02:40 <clandmeter> barthalion: why did you need those repo's?
2015-06-28 21:04:13 <barthalion> clandmeter: I wanted to take a look at the code for my curiosity mostly
2015-06-28 21:05:04 <clandmeter> most probably not the best example to look at :)
2015-06-28 21:05:23 <pluesch> clandmeter: I'm currently playing with taports-turbo too ^^ ty for sharing :)
2015-06-28 21:05:31 <pluesch> aports-turbo*
2015-06-28 21:06:15 <clandmeter> yw
2015-06-28 21:06:21 <clandmeter> patches welcome :)
2015-06-28 21:07:07 <clandmeter> i never tried lua before, so there is enough to improve.
2015-06-28 21:07:59 <clandmeter> k0r10n: im trying to build chromium now. takes ages though...
2015-06-28 23:48:06 <vkris> ok, iftop issue solved, my mistake
2015-06-28 23:48:56 <vkris> was there someone saying , they had some pkgs showing in /etc/apk/world, but it did not get installed in next reboot !?
2015-06-28 23:49:57 <vkris> I found a workaround, add "apk fix" to /etc/local.d/apkfix.start file, chmod +x it, and voila!!
2015-06-28 23:51:27 <vkris> I think I might me running my first AL on bare metal sooonish ;)
2015-06-28 23:59:19 <vkris> and did I mention, "Al box with custom made chassis" ;-))
2015-06-29 01:23:16 <mitchty> kaniini is this kinda what you are looking for? https://github.com/mitchty/alpine-linux-ghc-bootstrap/tree/apk I'm not sure how you can get from bootstrap compiler to a self hosting apk within the apkfile itself
2015-06-29 01:24:49 <mitchty> so what i setup was a makefile that runs the bootstrap stuff, uses docker to yoink out the cross compiler, then build ghc/ghc-dev/ghc-doc packages that you can then use to self host a "real" apk that depends on those apks
2015-06-29 01:26:51 <mitchty> missing the equivalent of stage1-2 equivalent of cross compiler compiling for apks
2015-06-29 01:28:43 <mitchty> or do you mean the apkbuild does both bootstrap and non?
2015-06-29 01:29:38 <mitchty> with say a BOOTSTRAP=/tar.xz abuild -r as an example
2015-06-29 01:32:28 <mitchty> (note, still building that atm so its not 100%)
2015-06-29 02:00:13 <ahills> clandmeter: yeah it takes hours on my laptop
2015-06-29 02:01:36 <ahills> vkris: pics of the chassis?
2015-06-29 03:34:02 <kaniini>  what I am saying is
2015-06-29 03:34:12 <kaniini> you have docker stuff to build the bootstrap
2015-06-29 03:34:36 <kaniini> we have an apkbuild which installs from the output files of that
2015-06-29 03:35:07 <kaniini> and then the real ghc depends on ghc-bootstrap
2015-06-29 03:35:47 <kaniini> but we can do better using a trick with provides and versions since apk will prefer the higher version
2015-06-29 03:35:54 <kaniini> 
ACTION pondera

2015-06-29 03:35:58 <kaniini> ponders*
2015-06-29 03:40:29 <mitchty> well i don't see a point in depending on ghc-bootstrap after you're built
2015-06-29 03:40:43 <mitchty> in that you can build ghc with itself after that point
2015-06-29 03:41:40 <mitchty> i.e the bootstrap is a one time thing, vs a constant time thing, even then ghc generally doesn't build with much past the prior versions of itself
2015-06-29 03:42:43 <mitchty> so it'll take a while for this to finish but that branch has APKBUILDS that do bootstrap/non
2015-06-29 03:42:58 <mitchty> all under the ghc/cabal-install apkbuilds
2015-06-29 03:43:19 <mitchty> i might've missed something but the goal is in the end apk files of the finalized packages
2015-06-29 03:43:45 <mitchty> the bootstrap stuff only gets used to generate the final apks
2015-06-29 03:45:35 <mitchty> i can see keeping the bootstrap stuff working just in case or to help new arch porting, but past that its somewhat of a dead weight
2015-06-29 04:03:42 <mitchty> anyway, you know apkbuilds better than i do, so if you could fork that branch and show me what it is you want to do it would help, cause either I'm missing something, dense, or both probably
2015-06-29 08:08:57 <leo-unglaub> hey guyes :)
2015-06-29 08:09:38 <leo-unglaub> ncopa: did you get the patches from me?
2015-06-29 08:14:15 <ncopa> hi
2015-06-29 08:15:51 <ncopa> this one? http://git.alpinelinux.org/cgit/aports/commit/?id=b86e6a4d7aca9e2559b822d3c06d5ac850f6db9c
2015-06-29 08:22:23 <ncopa> fabled: blocker for v3.2.1: clang fails on v3.2: http://sprunge.us/PfOO
2015-06-29 08:22:52 <ncopa> it was introduced with #4235
2015-06-29 08:23:33 <ncopa> i pushed the clang fixes to v3.2 post v3.2.0 release
2015-06-29 08:24:18 <ncopa> http://git.alpinelinux.org/cgit/aports/commit/?h=3.2-stable&id=a1d93256fc622fc25e4379c261c744957f387b5f
2015-06-29 08:27:49 <leo-unglaub> ncopa: ah, great :) an the other one?
2015-06-29 08:28:08 <leo-unglaub> i have to fix git here, because i could not send the, via git here
2015-06-29 08:28:22 <leo-unglaub> but i hope the patches where correct in general
2015-06-29 08:28:23 <leo-unglaub> :)
2015-06-29 08:29:20 <ncopa> the patches looks correct
2015-06-29 08:29:32 <ncopa> why is suid bit needed?
2015-06-29 08:30:39 <leo-unglaub> this script is spawned as user root and then changes down to the user of the vhost. this is done so that every vhost can run as a complete different user
2015-06-29 08:31:09 <ncopa> ok
2015-06-29 08:31:21 <ncopa> https://www.hiawatha-webserver.org/
2015-06-29 08:31:23 <ncopa> :)
2015-06-29 08:31:59 <leo-unglaub> i have read the c code very often that is used with suid and its clean
2015-06-29 08:32:06 <ncopa> i think it is kinda cool that the 3 goas are: secure, easy to use (simple), lightweight (small)
2015-06-29 08:32:13 <leo-unglaub> hiawatha in general is a very awesome webserver
2015-06-29 08:32:29 <ncopa> it seems like they have similar goals as alpine :)
2015-06-29 08:32:39 <leo-unglaub> hahahah, yeah .... it should seam familiar *g*
2015-06-29 08:32:56 <ncopa> they even have mountains on the their front page
2015-06-29 08:33:09 <leo-unglaub> *g*
2015-06-29 08:33:17 <kaniini> what is with webservers being named after native american tribes/people/whatever
2015-06-29 08:33:32 <leo-unglaub> if you read this section here: https://www.hiawatha-webserver.org/manpages/hiawatha you will never use another webserver
2015-06-29 08:33:37 <kaniini> oh, well, apache did it hurr hurr hurr
2015-06-29 08:33:43 <leo-unglaub> even the ban settings alone are so awesome
2015-06-29 08:34:08 <kaniini> you're aware other webservers can do that right
2015-06-29 08:34:17 <kaniini> mod_security, naxsi, etc.
2015-06-29 08:34:26 <leo-unglaub> kaniini: not even close
2015-06-29 08:34:38 <leo-unglaub> apache's mod_security is horrible
2015-06-29 08:35:58 <kaniini> you know the webserver that is really shitty?
2015-06-29 08:36:00 <kaniini> nginx
2015-06-29 08:36:11 <kaniini> i tried to patch nginx to log to a pipe
2015-06-29 08:36:20 <kaniini> and the code was so awful i gave up
2015-06-29 08:36:37 <kaniini> the logging in nginx is literally all over the place
2015-06-29 08:37:03 <leo-unglaub> yes, i know. ngnix is fast, but not clean at all
2015-06-29 08:37:10 <leo-unglaub> thats why i use hiawatha for years
2015-06-29 08:37:12 <leo-unglaub> :)
2015-06-29 08:37:19 <ncopa> leo-unglaub: can you please join #alpine-commits?
2015-06-29 08:37:25 <leo-unglaub> ncopa: this is the suid code: https://github.com/hsleisink/hiawatha/blob/master/src/cgi-wrapper.c
2015-06-29 08:37:30 <leo-unglaub> i have audited it twice
2015-06-29 08:46:16 <leo-unglaub> i have antother patch locally that removes gstreamer from pidgin
2015-06-29 08:46:28 <leo-unglaub> A: pidgin uses the outdates gstreamer 0.1
2015-06-29 08:46:38 <leo-unglaub> B: audio/video calls never worked in pidgin
2015-06-29 08:46:50 <leo-unglaub> it makes everything smaller and less bloaded, interrested?
2015-06-29 08:47:17 <ncopa> hm
2015-06-29 08:47:38 <ncopa> so, pidgin does not work with audio/video?
2015-06-29 08:47:52 <ncopa> i never made it work with it, but i kinda hoped that it would work some day
2015-06-29 08:48:09 <leo-unglaub> in theory it does, but in my 8 years using it i never got it to work
2015-06-29 08:49:47 <ncopa> :)
2015-06-29 08:51:33 <ncopa> that pretty much correspond with my experience too
2015-06-29 08:51:43 <leo-unglaub> this is my configure for it locally: https://paste.debian.net/plain/265303
2015-06-29 08:51:58 <barthalion> I don't remember using it either, so lgtm
2015-06-29 08:52:13 <leo-unglaub> maybe there are some options you would love to put in the main package
2015-06-29 08:54:15 <ncopa> avahi
2015-06-29 08:54:21 <ncopa> whats avahi used for in pidgin?
2015-06-29 08:54:35 <ncopa> so you can do local net chat without server or similar?
2015-06-29 08:54:53 <kaniini> idk
2015-06-29 08:55:07 <kaniini> probably for the non-working a/v stuff
2015-06-29 08:55:50 <ncopa> so we disable gstreamer, farstream vv?
2015-06-29 08:58:46 <ncopa> do we want disable the dbus interface too?
2015-06-29 08:59:06 <kaniini> probably not
2015-06-29 09:00:17 <barthalion> avahi seems to be just for services discovery in local network
2015-06-29 09:00:33 <kaniini> yes but iChat A/V
2015-06-29 09:00:40 <kaniini> was built ontop of bonjour
2015-06-29 09:00:46 <kaniini> so it is probably related to that
2015-06-29 09:10:47 <leo-unglaub> ncopa: i dont use dbus on pidgin because i hate dbus
2015-06-29 09:11:08 <leo-unglaub> however, they have massive dbus suport. right to the smallest detail. it's very impressive that they did that do good
2015-06-29 09:12:09 <leo-unglaub> avahi would be usefull for bonjour chatting, however, it does not work currently
2015-06-29 09:12:26 <leo-unglaub> Unable to establish connection with the local mDNS server.  Is it running?
2015-06-29 09:12:34 <leo-unglaub> but there is no way of getting that fixed
2015-06-29 09:16:29 <ncopa> somethign like this: http://sprunge.us/OdGK
2015-06-29 09:16:52 <ncopa> actually
2015-06-29 09:17:07 <ncopa> we could split things into subpackages i think
2015-06-29 09:18:48 <leo-unglaub> i doubt that, because gstreamer is compiled into the main binary
2015-06-29 09:18:55 <leo-unglaub> and no .so file
2015-06-29 09:18:58 <ncopa> except gstream indeed
2015-06-29 09:19:15 <ncopa> i was thinking avahi and dbus support
2015-06-29 09:19:15 <leo-unglaub> also dbus is not a library as far as i remember
2015-06-29 09:19:48 <ncopa> ok
2015-06-29 09:19:50 <ncopa> dbus too
2015-06-29 09:19:57 <ncopa> is compiled into pidgin
2015-06-29 09:20:28 <leo-unglaub> the problem with dbus is, that pidgin puts sooo much on the local channel that it actually causes a lot of load
2015-06-29 09:20:38 <leo-unglaub> because every keystroke is a dbus user typing event
2015-06-29 09:20:43 <ncopa> ugh
2015-06-29 09:21:02 <leo-unglaub> you can do everything over dbus in pidgin, that impressive
2015-06-29 09:21:13 <leo-unglaub> but also expensive, because every status change, ... is send to dbus
2015-06-29 09:21:23 <leo-unglaub> and if you are in 10+ irc channels, thats a lot of stuff
2015-06-29 09:21:39 <ncopa> but does anyone use pidgin with dbus though?
2015-06-29 09:21:55 <leo-unglaub> it gets even worse if you keep in mind that dbus uses string copys instead of handing pointers around
2015-06-29 09:22:30 <leo-unglaub> the only real usecase for pidgin with dbus is that in thunar/caja you can rightclick a file and send it to a jabber buddy
2015-06-29 09:22:40 <leo-unglaub> the contact list there is fetched via dbus if i remember correctly
2015-06-29 09:22:49 <leo-unglaub> other that that, it's not used to my knowlage
2015-06-29 09:23:46 <ncopa> ok
2015-06-29 09:24:08 <ncopa> i understand your 10+ irc channel issue
2015-06-29 09:24:29 <ncopa> i wonder if the longterm solution is to add an option to irc module: "Disable dbus"
2015-06-29 09:24:31 <ncopa> or similar
2015-06-29 09:25:05 <leo-unglaub> i think the longtime solution is to wait for the message bus that is coming in the linux kernel
2015-06-29 09:25:14 <ncopa> heh :)
2015-06-29 09:25:20 <leo-unglaub> if they do it better and faster than dbus hopefully they will all migrate
2015-06-29 09:26:04 <crow> what about gajim, should work for video/audio and have python OTR plugin
2015-06-29 09:26:18 <ncopa> crow: does it actually work?
2015-06-29 09:26:50 <ncopa> ah gajim != pidgin
2015-06-29 09:35:49 <ncopa> ok
2015-06-29 09:36:58 <ncopa> we coudl split out libpurple-bonjour, libpurple-icq (or oscar or what we call it), libpurple-xmpp, libpurple-yahoo
2015-06-29 09:37:20 <ncopa> at least libpurple-bonjoir would avoid pull in extra avahi dep
2015-06-29 09:37:36 <crow> ncopa gajim should work,
2015-06-29 09:37:45 <leo-unglaub> i like that
2015-06-29 09:37:57 <leo-unglaub> it would make it smaller again :)
2015-06-29 09:37:59 <leo-unglaub> small is good :)
2015-06-29 09:38:47 <ncopa> we already split out the text ui, finch
2015-06-29 09:39:09 <ncopa> im slightly in doubt what to do wht dbus
2015-06-29 09:40:11 <leo-unglaub> hmmm, have you ever used dbus in any application?
2015-06-29 09:40:36 <ncopa> yes. bluetooth
2015-06-29 09:40:45 <ncopa> for making bluez work
2015-06-29 09:40:55 <ncopa> but not with pidgin
2015-06-29 09:41:04 <leo-unglaub> bluez????
2015-06-29 09:41:06 <leo-unglaub> omg
2015-06-29 09:41:12 <leo-unglaub> that is anotherr sik application *g*
2015-06-29 09:41:17 <ncopa> :)
2015-06-29 09:41:39 <ncopa> there is no other way to get a bt keyboard work
2015-06-29 09:41:57 <ncopa> i have an apple keyboard
2015-06-29 09:42:09 <leo-unglaub> bluez is soo bad, even libreoffice looks at it and says: guys, take a little bit more care *g*
2015-06-29 09:42:10 <leo-unglaub> hahah
2015-06-29 09:42:20 <ncopa> :)
2015-06-29 09:42:48 <leo-unglaub> i would say remove dbus support until someon cryes
2015-06-29 09:43:29 <ncopa> thats what i'm thinking too
2015-06-29 09:43:33 <leo-unglaub> anther thing, have you every thought about switching from openssl to libressl?
2015-06-29 09:43:41 <ncopa> hehe
2015-06-29 09:43:51 <ncopa> we have discussed it
2015-06-29 09:45:09 <ncopa> i think the conclusion was something like: is it worth replacing something broken with something else broken
2015-06-29 09:45:16 <ncopa> even if it is slightly less broken
2015-06-29 09:45:20 <leo-unglaub> i think both suck compared to mbsdtls, but the huge benefit of libressl is the new api
2015-06-29 09:45:46 <leo-unglaub> because libssl is horrible to use for developers while libtls actually has an improved syntax
2015-06-29 09:45:54 <leo-unglaub> just from a developers point of view
2015-06-29 09:47:10 <ncopa> wouldnt all the apps using libssl need to be ported?
2015-06-29 09:47:23 <leo-unglaub> no, they ship both versions
2015-06-29 09:47:27 <ncopa> or is it enough with a simple rebuild
2015-06-29 09:47:47 <leo-unglaub> libressl is complete drop in replacement
2015-06-29 09:48:15 <leo-unglaub> the new api is just for newley developed applications
2015-06-29 09:48:23 <leo-unglaub> i dont know if you have ever used libssl
2015-06-29 09:48:28 <leo-unglaub> but it was impossible to get right
2015-06-29 09:48:48 <ncopa> i have heard that the api is not so good
2015-06-29 09:49:10 <leo-unglaub> http://www.openbsd.org/cgi-bin/man.cgi/OpenBSD-current/man3/tls_accept_fds.3?query=tls_init&sec=3
2015-06-29 09:49:28 <leo-unglaub> this is the new one, if you compare it to the old one it's like .... awesome
2015-06-29 09:49:49 <leo-unglaub> http://www.openbsd.org/cgi-bin/man.cgi/OpenBSD-current/man3/ssl.3?query=SSL&sec=3
2015-06-29 09:49:56 <leo-unglaub> if you compare them you see the difference *g*
2015-06-29 10:01:24 <leo-unglaub> i have to go, i have to clean up my apartment, because i get visitors in 6 hours and it looks here like a server rack exploded ....
2015-06-29 10:01:28 <leo-unglaub> see you later :)
2015-06-29 10:01:44 <kaniini> it is not 'awesome'
2015-06-29 10:02:01 <kaniini> libressl's new libtls is not properly thread safe
2015-06-29 10:02:37 <leo-unglaub> kaniini: what?
2015-06-29 10:02:41 <leo-unglaub> since when?
2015-06-29 10:02:58 <kaniini> since always
2015-06-29 10:03:43 <kaniini> it has initialization code but does not provide explicit init/fini functions
2015-06-29 10:04:03 <kaniini> instead depending on [[constructor]] attributes
2015-06-29 10:04:07 <kaniini> and other uglyness
2015-06-29 10:04:39 <kaniini> on top of that, they have removed entire pieces of the codebase not understanding what they did, and that resulted in serious regressions (admittedly they were fixed quickly, but...)
2015-06-29 10:05:05 <kaniini> libressl is no answer, fixing programs to use a better TLS stack is the answer
2015-06-29 10:05:36 <leo-unglaub> libtls is as thread safe as the openssl counterpart
2015-06-29 10:05:59 <leo-unglaub> like describes here: https://wiki.openssl.org/index.php/Libcrypto_API#Thread_Safety
2015-06-29 10:06:13 <leo-unglaub> libtls takes a little bit of that pain away
2015-06-29 10:06:14 <kaniini> you do not get to compare something new to the old thing when you are using it as a perceived advantage over the old thing, sorry
2015-06-29 10:06:40 <leo-unglaub> i did say, that is is not perfect, but if compared 1000 times better
2015-06-29 10:06:51 <kaniini> where do you draw that conclusion
2015-06-29 10:07:09 <kaniini> just because something is "simpler" doesn't mean it is more correct
2015-06-29 10:07:20 <kaniini> in security software, correctness is what matters more than anything else
2015-06-29 10:07:35 <leo-unglaub> correct, i never sayed that
2015-06-29 10:07:46 <kaniini> then stop boosting it
2015-06-29 10:08:07 <leo-unglaub> well, i provide arguments for it
2015-06-29 10:08:22 <leo-unglaub> with is something i did not see from you
2015-06-29 10:08:30 <leo-unglaub> you only had your opinion
2015-06-29 10:09:12 <kaniini> facts are not opinions.  libtls solves no real-world problem with openssl api other than providing a new API.
2015-06-29 10:09:24 <kaniini> it is still the same garbage in the backend
2015-06-29 10:09:31 <leo-unglaub> if you take the statistics for it. from the last 20 openssl security problems only 7 existed in libressl
2015-06-29 10:09:36 <kaniini> if we're moving to a new API, then we should get rid of the garbage too.
2015-06-29 10:09:56 <leo-unglaub> kaniini: i think you dont understand the problem
2015-06-29 10:10:04 <leo-unglaub> no one is moving to the new api for existing applications
2015-06-29 10:10:08 <kaniini> the problem is you.
2015-06-29 10:10:27 <leo-unglaub> because you would have to patch tousands of programs
2015-06-29 10:10:34 <leo-unglaub> and thats not going to happen
2015-06-29 10:10:56 <leo-unglaub> or do you see php, apache, tor, pidgin, ... all switch to an new ssl lib in the next year?
2015-06-29 10:11:01 <leo-unglaub> they are not going to do that
2015-06-29 10:11:14 <leo-unglaub> so all libressl tryes to do is to improve the backend for that
2015-06-29 10:11:23 <leo-unglaub> and as i sayed above they did
2015-06-29 10:11:24 <kaniini> i'm glad that you have great ideas on what alpine should do, but we're not going to just abruptly change the SSL library because openbsd fanboys love it so much
2015-06-29 10:11:43 <kaniini> "from the last 20 openssl security problems only 7 existed in libressl" is largely because they removed large amounts of functionality
2015-06-29 10:12:28 <leo-unglaub> see, thats where you are wrong. they did not remove actual functions from the library, they removed for example old and usess compatibility stuff
2015-06-29 10:12:35 <leo-unglaub> like the internal own memory mapper
2015-06-29 10:12:39 <leo-unglaub> that has no reason to be there
2015-06-29 10:12:49 <kaniini> no i am not wrong, they removed several AEADs
2015-06-29 10:12:55 <kaniini> you have no idea what you're talking about
2015-06-29 10:13:20 <leo-unglaub> give me an example of something that they removed that is still relevant for todays crypto?
2015-06-29 10:13:44 <kaniini> who decides what is 'still relevant for todays crypto'?  theo de raadt?
2015-06-29 10:13:59 <leo-unglaub> no, research does
2015-06-29 10:14:00 <kaniini> then there's the PRNG stuff they did which was way broken on linux
2015-06-29 10:14:25 <kaniini> how many actual cryptographers are on the openbsd team?  oh, right, zero.
2015-06-29 10:14:58 <bCQc0P3DIqbh> EADS are very much relevant for todays crypto
2015-06-29 10:14:58 <leo-unglaub> what is an "actual cryptographer" ?
2015-06-29 10:15:07 <leo-unglaub> dont say NIST people
2015-06-29 10:15:09 <bCQc0P3DIqbh> they really removed some of those?
2015-06-29 10:15:21 <kaniini> leo-unglaub: someone who has specialized cryptography degree
2015-06-29 10:16:01 <kaniini> we could replace openssl's own memory functions with intrinsics too
2015-06-29 10:16:08 <kaniini> this isn't archlinux
2015-06-29 10:16:40 <kaniini> the people who bring you alpine have serious dev experience and have no problems getting down in the trenches
2015-06-29 10:17:15 <leo-unglaub> right ... sorry i for forgott that i am only a "openbsd fanboy"
2015-06-29 10:17:42 <leo-unglaub> no, you got me, i have only googled something about libressl and wanted to start a discussion about it
2015-06-29 10:17:56 <kaniini> yes, and we are tired of having the discussion
2015-06-29 10:18:28 <kaniini> at next ABI break, we will re-evaluate if libressl is a good choice
2015-06-29 10:18:41 <kaniini> until then, our openssl is already patched in various ways anyway
2015-06-29 10:19:16 <kaniini> alpine openssl already uses memory intrinsics
2015-06-29 10:19:34 <kaniini> we did that before libressl was even a thing
2015-06-29 10:20:22 <vegai> kaniini: hey! :P
2015-06-29 10:21:06 <barthalion> vegai: sup
2015-06-29 10:21:13 <kaniini> so i think that is what you're not understanding -- to us, libressl is basically equivilant to what we're shipping right now, except with ABI/API breaks (OPENSSL_VERSION_NUMBER == 0x2000000)
2015-06-29 10:21:17 <vegai> barthalion: howdy
2015-06-29 10:22:01 <leo-unglaub> kaniini: thats where i disagree with you. saying that your patches do the same as libressl is simpy not correct
2015-06-29 10:22:44 <vegai> this isn't arch linux -- arch linux has users
2015-06-29 10:22:46 <vegai> 
ACTION dodges

2015-06-29 10:23:00 <kaniini> leo-unglaub: i have no doubt they have done more digging through the crap pile than we have
2015-06-29 10:23:08 <kaniini> leo-unglaub: the problem is
2015-06-29 10:23:17 <kaniini> leo-unglaub: is that digging worth justifying an API/ABI break
2015-06-29 10:23:40 <kaniini> leo-unglaub: for now, we decided no.  there will be another discussion when openssl ABI breaks again.
2015-06-29 10:23:45 <barthalion> vegai: if only numbers == quality :P
2015-06-29 10:24:13 <barthalion> but I disgress
2015-06-29 10:24:35 <kaniini> vegai: teenagers running compiz and thinking they're leet isn't really the userbase we're going for
2015-06-29 10:24:38 <kaniini> :)
2015-06-29 10:24:51 <leo-unglaub> in some cases you are right, that just a cleanup is not worth breaking existing stuff. but in this case 99% of all packages would not be accected
2015-06-29 10:25:00 <kaniini> leo-unglaub: that is wrong
2015-06-29 10:25:10 <kaniini> leo-unglaub: the SONAME on libressl libssl is libssl.so.2.0
2015-06-29 10:25:20 <kaniini> leo-unglaub: which means every package has to be rebuilt
2015-06-29 10:25:38 <kaniini> could it be done -- sure
2015-06-29 10:25:44 <kaniini> is it worth doing -- we decided no
2015-06-29 10:25:56 <vegai> :)
2015-06-29 10:26:02 <kaniini> admittedly this decision was done early on
2015-06-29 10:26:10 <kaniini> when libressl did not have much of a roadmap or plans
2015-06-29 10:26:17 <kaniini> and we should reevaluate against 2.1
2015-06-29 10:26:42 <leo-unglaub> on this speciffic issue i have to pass, because i dont have the build folder open right now, but as far as i remember you can compile it as a drop in replacement
2015-06-29 10:26:56 <leo-unglaub> and they did break API ..
2015-06-29 10:27:06 <leo-unglaub> but one that is not used on linux
2015-06-29 10:27:13 <kaniini> well, my main issue with their API breaks is
2015-06-29 10:27:17 <leo-unglaub> the only argument where i have to agree with you is the PRNG issue
2015-06-29 10:27:19 <kaniini> OPENSSL_VERSION_NUMBER = 0x2000000
2015-06-29 10:27:28 <kaniini> they should have left this alone
2015-06-29 10:27:44 <kaniini> and any libressl in alpine would be patched to revert that stupidity
2015-06-29 10:27:44 <leo-unglaub> they focus so much on there arc4random syscall that it is a pain to use on linux
2015-06-29 10:28:02 <leo-unglaub> but thats the only thing they did wrong .. and even fixed it in there export version
2015-06-29 10:28:18 <kaniini> that is not the only thing they did wrong
2015-06-29 10:28:31 <kaniini> changing OPENSSL_VERSION_NUMBER is a broken change
2015-06-29 10:30:09 <kaniini> the reason why it is wrong is because now programs which want to use libssl 1.0.2 features
2015-06-29 10:30:13 <kaniini> cannot on libressl
2015-06-29 10:30:40 <kaniini> so for something that claims to be "compatible with the openssl 1.0.1g API" it is not there
2015-06-29 10:34:32 <kaniini> leo-unglaub: i also think they plan on removing support for cryptographic accelerator cards
2015-06-29 10:34:48 <leo-unglaub> i dont get your problem with OPENSSL_VERSION_NUMBER
2015-06-29 10:35:04 <leo-unglaub> they use the same for LIBRESSL_VERSION_NUMBER
2015-06-29 10:35:09 <leo-unglaub> to be the exact same
2015-06-29 10:36:16 <kaniini> because it breaks
2015-06-29 10:36:25 <kaniini> #ifdef OPENSSL_VERSION_NUMBER > 0x1000100
2015-06-29 10:36:40 <kaniini>     ... openssl 1.0.2-using code here ...
2015-06-29 10:36:43 <kaniini> #else
2015-06-29 10:36:56 <kaniini>     ... openssl <= 1.0.1 code here
2015-06-29 10:36:59 <kaniini> #endif
2015-06-29 10:37:06 <kaniini> that #ifdef should be #if but w/e
2015-06-29 10:37:16 <kaniini> why does it break
2015-06-29 10:37:24 <kaniini> because libressl does not provide the new 1.0.2 apis
2015-06-29 10:37:30 <kaniini> but it claims to be 2.0.0
2015-06-29 10:37:39 <kaniini> so now devs have to do something like this
2015-06-29 10:38:05 <kaniini> #if OPENSSL_VERSION_NUMBER > 0x1000100 && !defined(LIBRESSL_VERSION_NUMBER)
2015-06-29 10:38:20 <kaniini> why should they have to be concerned about this incompatibility
2015-06-29 10:38:51 <leo-unglaub> i still dont get what should be missing from libressl that you even have to do this check
2015-06-29 10:39:13 <kaniini> leo-unglaub: TLS_server_method()
2015-06-29 10:39:19 <leo-unglaub> 2.2.0 from june 11?
2015-06-29 10:39:31 <kaniini> who cares
2015-06-29 10:39:38 <kaniini> that doesn't cover 2.0 or 2.1
2015-06-29 10:39:41 <kaniini> which do not have it
2015-06-29 10:40:01 <kaniini> "use 2.2.0" is not an acceptable answer to most people who do not want to compile a new version of their TLS library by hand
2015-06-29 10:41:13 <kaniini> so when they stop writing cheques they cannot cash (claiming to support all APIs of openssl 1.0.1 .. 1.99.99), perhaps then we will not use that as a negative against migrating
2015-06-29 10:41:51 <kaniini> when they put back ENGINE support so we can take advantage of padlock, aes-ni, etc, perhaps we will consider too
2015-06-29 10:42:18 <kaniini> from my perspective we lose a lot for the gain of memory intrinsics and admittedly a lot of code auditing
2015-06-29 10:42:29 <kaniini> we've put the intrinsics in our own openssl build
2015-06-29 10:42:40 <kaniini> and maybe someday will port over more patches
2015-06-29 10:42:58 <kaniini> but losing ENGINE support, this OPENSSL_VERSION_NUMBER lie, etc.
2015-06-29 10:43:02 <kaniini> is really not good
2015-06-29 10:43:21 <kaniini> there are a lot of people using alpine + crypto acceleration for VPN servers
2015-06-29 10:43:31 <kaniini> if we switch to libressl they will lose their acceleration
2015-06-29 10:43:57 <kaniini> so i mean, again, it's great that the openbsd guys made libressl
2015-06-29 10:44:07 <kaniini> but they have removed things we do support in alpine
2015-06-29 10:45:56 <leo-unglaub> i just had to look it up, libressl has TLS_server_method
2015-06-29 10:46:07 <kaniini> yes, in 2.2
2015-06-29 10:46:19 <kaniini> again, 2.0 and 2.1 do not have it.
2015-06-29 10:46:29 <leo-unglaub> but the method name is prefixed for some reason with TLSv1_1_server_method
2015-06-29 10:46:37 <kaniini> ...
2015-06-29 10:46:50 <kaniini> TLS_server_method() and TLSv1_1_server_method() are different
2015-06-29 10:47:02 <kaniini> TLSv1_1_server_method() is TLSv1.1 only
2015-06-29 10:47:04 <leo-unglaub> so i guess i have to give you that one, you actually found a bug in libressl
2015-06-29 10:47:18 <leo-unglaub> yes, it exists for every tls version seperatly
2015-06-29 10:47:38 <leo-unglaub> unless they pull some magic crap that maps the call in some way
2015-06-29 10:47:39 <kaniini> TLS_server_method() allows any TLS version to be negotiated
2015-06-29 10:47:46 <kaniini> anyway
2015-06-29 10:48:00 <kaniini> all of this is stupid, there's so many reasons to keep things the way they are
2015-06-29 10:48:24 <kaniini> openssl ripped out (or, at least, did something to) cryptographic acceleration engine support
2015-06-29 10:48:38 <kaniini> will it effect alpine users who depend on it
2015-06-29 10:48:49 <kaniini> again: this isn't arch linux
2015-06-29 10:49:00 <kaniini> whimsically breaking people's stuff is not really what we're trying to do here
2015-06-29 10:49:05 <leo-unglaub> why do you always come back to arch linux?
2015-06-29 10:49:18 <leo-unglaub> i dont see the connection to it
2015-06-29 10:49:40 <kaniini> i am just using them as an example of a distribution which has historically had no qualms with breaking people's setups
2015-06-29 10:51:12 <leo-unglaub> i have no idea why you think i want to break peoples setup
2015-06-29 10:51:40 <kaniini> well, you are saying "lets use libressl!!!!oneoneone" and like everyone else who has said this, you've not provided anything resembling a plan on how to get there
2015-06-29 10:51:54 <kaniini> nor what actual real-world impact such a major decision would cause
2015-06-29 10:52:06 <kaniini> you just say, like the others, "oh, 99% of packages will rebuild no problem"
2015-06-29 10:52:07 <leo-unglaub> actually i have a plan on how to get there ...
2015-06-29 10:52:18 <leo-unglaub> but you called me a "fanboy" before i could even get to it
2015-06-29 10:52:38 <leo-unglaub> the fanboy thing is specially funny because i hate openbsd on laptops or servers because it's fucking slow as hell
2015-06-29 10:52:41 <kaniini> oh, getting there is the easy part
2015-06-29 10:53:05 <kaniini> it's the actual impact estimation which is always "oh, 99% of packages will rebuild no problem" which demonstrates a fundamental misunderstanding of the problem
2015-06-29 10:53:11 <barthalion> well, yeah, because getting stuck with something for clean upgrade path is something we should desire…
2015-06-29 10:53:24 <barthalion> and Arch switched to libressl…
2015-06-29 10:53:51 <barthalion> I'm sometimes really tired of Alpine to Arch comparison, completely different distributions
2015-06-29 10:53:51 <kaniini> good for arch!  go use arch!
2015-06-29 10:53:55 <barthalion> apples to oranges
2015-06-29 10:54:23 <barthalion> I'm being sarcastic now
2015-06-29 10:54:34 <kaniini> my comparison wasn't comparing alpine to arch
2015-06-29 10:54:35 <barthalion> we didn't switch to libressl for the very same reason Alpine did not
2015-06-29 10:55:00 <barthalion> "this isn't arch" looks like one =)
2015-06-29 10:55:05 <kaniini> my statement was just that arch has not historically had problems making changes which caused massive breakage
2015-06-29 10:55:21 <kaniini> and yes, a clean upgrade path *is* a goal of alpine
2015-06-29 10:55:36 <leo-unglaub> kaniini: well, ncopa just sayed it before: "lets remove dbus from it until someone cryes and wants it back"
2015-06-29 10:55:52 <kaniini> that is for desktop
2015-06-29 10:55:54 <barthalion> I didn't know trolling is another goal either
2015-06-29 10:55:58 <leo-unglaub> so saying "omg never break anything" is not the solution either
2015-06-29 10:55:59 <barthalion> 
ACTION shrugs

2015-06-29 10:56:01 <barthalion> eot from my side'
2015-06-29 10:56:19 <leo-unglaub> i have some the test, my laptop runs on alpine + libressl
2015-06-29 10:56:30 <kaniini> great
2015-06-29 10:56:32 <leo-unglaub> and i had none of the problem you where talking about
2015-06-29 10:56:33 <kaniini> that's a nice test, it really is.
2015-06-29 10:56:59 <kaniini> will an alpine using sysadmin who has a 4000 connection vpn server on a via padlock system
2015-06-29 10:57:00 <leo-unglaub> also, stability guidelines should be the same for sever and desktop
2015-06-29 10:57:07 <kaniini> still be able to use his padlock acceleration
2015-06-29 10:57:30 <leo-unglaub> kaniini: i have no idea, thats why there should be tests before its done
2015-06-29 10:57:44 <leo-unglaub> you never heared me say: lets swithc to it and role it out
2015-06-29 10:57:49 <leo-unglaub> i would never even suggest that
2015-06-29 10:57:56 <leo-unglaub> but having a discussion about it is important
2015-06-29 10:58:01 <kaniini> welp, that's what literally everyone else before you said
2015-06-29 10:58:14 <leo-unglaub> well, i am not an idiot
2015-06-29 10:58:19 <leo-unglaub> i know what the impact is
2015-06-29 10:58:24 <barthalion> the point is that we had this discussion already
2015-06-29 10:58:26 <barthalion> 3 or 4 times
2015-06-29 10:58:26 <leo-unglaub> and thats why i tryed it on my evice before
2015-06-29 10:59:00 <kaniini> that libressl works on your laptop is great
2015-06-29 10:59:13 <kaniini> but it doesn't really paint a real picture of what real-world impact is
2015-06-29 10:59:18 <barthalion> so it's nice thing to have in future, but we can't afford switching around forks every time something new pops out
2015-06-29 10:59:37 <kaniini> yeah that's another concern that has been raised
2015-06-29 10:59:43 <kaniini> libressl guys have said numerous times
2015-06-29 10:59:45 <leo-unglaub> kaniini: of course its not the complete picture, i never sayed that
2015-06-29 10:59:48 <kaniini> if they run out of money it's the end of libressl
2015-06-29 10:59:55 <kaniini> so why would we want to switch to that
2015-06-29 11:01:06 <leo-unglaub> sorry, but that is simply not true, they have integratet it soooo deply in openbsd, that they cannot get it out anymore
2015-06-29 11:01:16 <kaniini> in openbsd, sure
2015-06-29 11:01:18 <leo-unglaub> because they rewrote stuff for the new api of libtls
2015-06-29 11:01:21 <kaniini> and libressl-portable?
2015-06-29 11:01:46 <leo-unglaub> basically the only difference there is one patch for the arc4random syscall
2015-06-29 11:02:00 <leo-unglaub> so saying they will stop it if they have no money left is simply wrong
2015-06-29 11:02:30 <kaniini> still, from a release engineering standpoint that type of wording
2015-06-29 11:02:33 <kaniini> makes us nervous about it
2015-06-29 11:02:59 <leo-unglaub> do you have that quote somewhere?
2015-06-29 11:03:11 <kaniini> dude.
2015-06-29 11:03:14 <leo-unglaub> because i would be interrested in who sayed that
2015-06-29 11:03:18 <kaniini> it was on libressl.org website
2015-06-29 11:03:21 <kaniini> portable is dependent on funding
2015-06-29 11:03:42 <leo-unglaub> well, thats something completely different than what you just sayed
2015-06-29 11:04:22 <kaniini> no it's not.  how is it different.
2015-06-29 11:04:31 <kaniini> if portable is dependent on funding,
2015-06-29 11:04:39 <kaniini> then if there is no funding, then there is no portable
2015-06-29 11:04:55 <kaniini> just because they shipped some releases doesn't mean that the funding requirement went away
2015-06-29 11:05:47 <kaniini> anyway
2015-06-29 11:05:50 <kaniini> leo-unglaub: if you really believe in libressl
2015-06-29 11:06:14 <kaniini> what you should do is write out a *really* *long* *email* explaining how/why we should make the jump
2015-06-29 11:06:23 <leo-unglaub> well the same argument is true for openssl, if they dont get fonding anymore, they dont develop it anymore
2015-06-29 11:06:26 <kaniini> what the migration plan will be
2015-06-29 11:06:27 <leo-unglaub> same with firefox, tor, ...
2015-06-29 11:06:32 <leo-unglaub> so what is your argument here?
2015-06-29 11:06:50 <kaniini> what real-world impacts there will be
2015-06-29 11:06:52 <barthalion> that we can live without firefox or tor…
2015-06-29 11:07:42 <leo-unglaub> i just thing its interresting that you say if libressl asks for funding its dangerous, and if openssl does it, its normal
2015-06-29 11:07:43 <ncopa> back from lunch
2015-06-29 11:07:49 <kaniini> what will happen to the customizations we have made to openssl (many of which are likely obsoleted by improvements in libressl, but...)
2015-06-29 11:07:51 <leo-unglaub> i dont get that kind of argumnt
2015-06-29 11:07:57 <ncopa> man its hot here, have you been figthing or something?
2015-06-29 11:08:27 <leo-unglaub> no, we have a discussion
2015-06-29 11:08:54 <kaniini> my concerns are basically this
2015-06-29 11:09:10 <kaniini> - how healthy is libressl vs. openssl project wise
2015-06-29 11:09:19 <kaniini> - what has happened to ENGINE support (crypto accel)
2015-06-29 11:09:22 <leo-unglaub> that is a valid argument
2015-06-29 11:10:01 <kaniini> - how will we fix the OPENSSL_VERSION_NUMBER lie?
2015-06-29 11:10:11 <ncopa> we did a serious evaluation of libressl not too long ago
2015-06-29 11:10:21 <ncopa> when the sec fix broke ABI in openssl
2015-06-29 11:10:47 <ncopa> (and it was an alpine developer who provided the fix for it too in the end)
2015-06-29 11:10:50 <kaniini> yes, and we decided to hold off still
2015-06-29 11:11:01 <clandmeter> if we evaluated it, please add it to www FAQ and be done with it.
2015-06-29 11:11:16 <ncopa> good idea
2015-06-29 11:11:21 <rnalrd> +1
2015-06-29 11:11:26 <kaniini> - what will happen with the other customizations we made to openssl?
2015-06-29 11:11:31 <kaniini> it's really those 4 big things
2015-06-29 11:11:58 <kaniini> there's probably other concerns fabled has
2015-06-29 11:12:17 <kaniini> but to have a serious discussion about libressl which has the potential in us moving on it
2015-06-29 11:12:23 <kaniini> those are the main areas that need to be answered
2015-06-29 11:12:29 <ncopa> leo-unglaub: we don't think openssl is superior libressl
2015-06-29 11:12:44 <ncopa> and there are things atleast I like with libressl
2015-06-29 11:13:03 <ncopa> but we concluded that it was not worth it atm
2015-06-29 11:13:11 <ncopa> might be in future, but not right now
2015-06-29 11:13:12 <kaniini> same, there are a lot of changes they've made to libressl that i agree with
2015-06-29 11:13:32 <kaniini> and, comparing stock openssl vs. libressl it's a no-brainer
2015-06-29 11:13:38 <kaniini> but alpine openssl has never really been stock
2015-06-29 11:13:43 <leo-unglaub> maybe you overread this, but i started it with: that both libressl and openssl are horrible
2015-06-29 11:13:57 <ncopa> then we agree :)
2015-06-29 11:13:57 <leo-unglaub> but my approach to it is if ijust get 10% moreseecurity, i take it
2015-06-29 11:14:12 <kaniini> the question is, do we get that 10% more security
2015-06-29 11:14:16 <kaniini> or is it really 1% more
2015-06-29 11:14:26 <ncopa> and how much work is it gain those 10%
2015-06-29 11:14:27 <kaniini> we already use memory intrinsics in our openssl
2015-06-29 11:14:33 <bCQc0P3DIqbh> well if you have 10% more security you still have n-10% side chans
2015-06-29 11:14:36 <leo-unglaub> and just from the stats, that is the case. again, from the last 20 security problems in openssl, only 7 existed in libressl
2015-06-29 11:14:51 <ncopa> we know that
2015-06-29 11:15:02 <ncopa> that was one of the thing i liked with libressl too
2015-06-29 11:15:03 <kaniini> the real question is: from the last 20 security problems in openssl, how many existed in alpine?
2015-06-29 11:15:18 <leo-unglaub> that is a good question
2015-06-29 11:15:42 <kaniini> we do not know, however, it is probably less than 20
2015-06-29 11:15:42 <barthalion> I guess grsec saves us a lot of times
2015-06-29 11:15:48 <ncopa> "too many" is probably the answer to that :)
2015-06-29 11:16:05 <kaniini> barthalion: even without grsec :)
2015-06-29 11:16:19 <ncopa> also, we do have libressl in edge/testing
2015-06-29 11:16:25 <ncopa> so people can test it
2015-06-29 11:16:38 <leo-unglaub> ncopa: my laptop is running on it (selfbuild)
2015-06-29 11:16:44 <ncopa> cool
2015-06-29 11:16:44 <leo-unglaub> thats why i am bringing it up
2015-06-29 11:17:03 <kaniini> is your laptop a 4000-user vpn server running on a server with a crypto accelerator card though
2015-06-29 11:17:06 <kaniini> ;)
2015-06-29 11:17:20 <leo-unglaub> kaniini: sure, but only when i am not using firefox *g*
2015-06-29 11:17:31 <ncopa> the padlock hw crypto is a requirement yes
2015-06-29 11:17:42 <leo-unglaub> google switches over to my laptop if they run out of bandwith *g*
2015-06-29 11:17:58 <ncopa> lol
2015-06-29 11:18:37 <ncopa> we do keep an eye on libressl
2015-06-29 11:18:38 <kaniini> padlock is an example because we've gone to extra effort to make padlock work with all aspects of the system
2015-06-29 11:18:51 <kaniini> there's other accelerators out there that we want to support at least in openssl
2015-06-29 11:19:07 <kaniini> and, we have ported fixes from libressl before
2015-06-29 11:22:36 <kaniini> leo-unglaub: regarding QA levels needing to be the same for server vs. desktop, the main reason why they're not is because literally 99% of alpine installs are on server
2015-06-29 11:23:06 <bCQc0P3DIqbh> allow me to ask how you define "10% more secure"? does that mean the effort to find a new bug takes 10% longer? or the exploit on the market is 10% more expensive? or the exploit takes 10% more time to bruteforce?
2015-06-29 11:23:22 <kaniini> that too
2015-06-29 11:23:47 <kaniini> presumably i think he means the effort required is 10% more difficult
2015-06-29 11:24:13 <ncopa> leo-unglaub: i am removing dbus support from pidgin. do you have suggestion for commit message?
2015-06-29 11:24:27 <ncopa> i want mention the 10+ irc channels and resouce issue
2015-06-29 11:24:49 <kaniini> wait
2015-06-29 11:24:56 <kaniini> people use pidgin as their irc client?
2015-06-29 11:24:56 <kaniini> dear god
2015-06-29 11:25:09 <leo-unglaub> i would argue with the fact, that basically no application uses the dbus support
2015-06-29 11:25:40 <ncopa> "no known application"
2015-06-29 11:25:58 <leo-unglaub> ncopa: yeah, exactly
2015-06-29 11:26:08 <leo-unglaub> some custom scripts prpobly do
2015-06-29 11:26:12 <kaniini> also, removing dbus support from pidgin is hardly the same thing as replacing the core SSL/TLS/Crypto framework
2015-06-29 11:26:27 <ncopa> nobody claim it is same thing
2015-06-29 11:26:34 <kaniini> he did earlier
2015-06-29 11:26:41 <leo-unglaub> kaniini: no, i did not.
2015-06-29 11:26:54 <leo-unglaub> i just argued that an api creak is an api break
2015-06-29 11:27:06 <kaniini> oh, ok
2015-06-29 11:27:18 <leo-unglaub> and that removing a feature is something other than dropping in an replacement with the same functionallity
2015-06-29 11:27:35 <kaniini> we do not know if it has the same functionality
2015-06-29 11:27:59 <ncopa> dropping functionality can be a good thing
2015-06-29 11:28:08 <leo-unglaub> yeah, i agree
2015-06-29 11:28:26 <leo-unglaub> i am glad that libressl for example dropped the entire visual basic compiler macro stuff in there
2015-06-29 11:28:30 <leo-unglaub> that was just horrible
2015-06-29 11:28:35 <leo-unglaub> and still is horrible in openssl
2015-06-29 11:28:51 <kaniini> yes, that's nice, but the OPENSSL_VERSION_NUMBER nonsense resulting in real-world breakage ( for example, https://github.com/atheme/charybdis/commit/1a4e224a4ed2f6e29299401d9029404359de21f7 ) is concerning
2015-06-29 11:28:56 <leo-unglaub> but kaniini is right about the engine support
2015-06-29 11:29:26 <kaniini> and engine has been mucked with
2015-06-29 11:29:39 <kaniini> whether or not that means the support is still there, i don't know
2015-06-29 11:29:44 <leo-unglaub> well, i am sure they fix that version issue if someone told them
2015-06-29 11:29:45 <kaniini> but /usr/lib/ssl/engines is gone
2015-06-29 11:30:03 <kaniini> well, maybe you should let them know
2015-06-29 11:30:05 <leo-unglaub> nah, as far as i know they removed it entirely to be readded later
2015-06-29 11:30:13 <kaniini> if you want to own this issue and make it happen
2015-06-29 11:30:20 <kaniini> this is the sort of stuff you're going to need to fix up
2015-06-29 11:30:32 <leo-unglaub> it's a small thing
2015-06-29 11:30:34 <kaniini> in meantime, we keep an eye on what they are doing and will port over things as needed
2015-06-29 11:30:48 <leo-unglaub> if thats the biggest problem, then we can be happy
2015-06-29 11:31:02 <kaniini> ENGINE being gone is by far a larger problem
2015-06-29 11:31:34 <kaniini> but both are problems which have impact
2015-06-29 11:33:40 <leo-unglaub> i am interrested to see if BoringSSL has the engine support
2015-06-29 11:33:55 <leo-unglaub> becasue as far as i know even they removed it do to beeing horrible insecure
2015-06-29 11:34:23 <kaniini> sure, but we have people with setups which are dependent on it
2015-06-29 11:34:32 <kaniini> is what i was saying
2015-06-29 11:34:33 <kaniini> i'm not saying ENGINE is great
2015-06-29 11:34:46 <kaniini> there are certainly some problems with it
2015-06-29 11:35:33 <leo-unglaub> yeah, but if even google replaces something because its horrible, then you know that it must have problems
2015-06-29 11:36:31 <leo-unglaub> during your libressl evaluation, have you looked at BoringSSL as well?
2015-06-29 11:36:35 <kaniini> if they have replaced it, then great
2015-06-29 11:36:39 <kaniini> if they removed it, that's not great
2015-06-29 11:37:03 <kaniini> there are tens of thousands of installs which are using padlock right now for vpn accel
2015-06-29 11:37:27 <kaniini> i'm not trying to be hard on you, i am just saying that we cannot break those people
2015-06-29 11:37:37 <kaniini> at least not right now
2015-06-29 11:37:59 <leo-unglaub> that depends, i tend to break stuff before i compromize a user
2015-06-29 11:38:02 <kaniini> we could say "18 months from now we get rid of padlock support because it's insecure", and that is something that would probably be ok
2015-06-29 11:38:07 <leo-unglaub> but thats my personal opinion
2015-06-29 11:38:14 <leo-unglaub> i am sure alpine has a different opinion
2015-06-29 11:39:06 <kaniini> the opinion is that alpine would have to deprecate the padlock support
2015-06-29 11:39:11 <kaniini> it's too widely used
2015-06-29 11:39:18 <ncopa> i dont think we have evaluated boringssl
2015-06-29 11:39:27 <kaniini> this is, again, something you would put in your thread on alpine-devel
2015-06-29 11:39:34 <kaniini> and this is why we haven't switched yet
2015-06-29 11:39:53 <kaniini> there are hard decisions that have to be made with consideration for our users
2015-06-29 11:39:56 <ncopa> i feel we have bigger fish ta catch
2015-06-29 11:40:26 <kaniini> if you want to try to make those decisions, write up a complete mail on alpine-devel and we can make a decision on it
2015-06-29 11:40:44 <kaniini> more depth is needed than what an irc conversation can provide
2015-06-29 11:41:28 <leo-unglaub> ncopa: BoringSSL is basically openssl forked by google
2015-06-29 11:41:34 <ncopa> i know
2015-06-29 11:41:39 <kaniini> we're aware of what it is
2015-06-29 11:41:49 <ncopa> i'm slightly sceptic to google
2015-06-29 11:42:14 <leo-unglaub> yeah, me to. i dont really trust google that much. thats why i did not bring it up before
2015-06-29 11:42:44 <kaniini> also they basically say don't use it if you depend on API/ABI stability: https://www.imperialviolet.org/2014/06/20/boringssl.html
2015-06-29 11:42:44 <leo-unglaub> specially if you keep in mind that they basically screwed over chrome users with there hidden audio extension
2015-06-29 11:43:36 <ncopa> "There are no guar­an­tees of API or ABI sta­bil­ity with this code: we are not aim­ing to re­place OpenSSL as an open-source pro­ject. "
2015-06-29 11:44:18 <leo-unglaub> yeah, sure
2015-06-29 11:44:18 <kaniini> leo-unglaub: what will probably happen is that we will probably do some mashup of libressl + openssl in the future
2015-06-29 11:44:38 <kaniini> we basically forked uclibc back in the day
2015-06-29 11:44:50 <kaniini> it's nothing we're incapable of doing :p
2015-06-29 11:45:42 <kaniini> i have a large list of things i would like to do with openssl given time to work on them
2015-06-29 11:45:58 <kaniini> getting rid of the broken and insecure ASM stuff is high on my list
2015-06-29 11:46:26 <kaniini> as is looking into making ENGINE more secure (but the argument against ENGINE is basically that one could snoop on the channel between the crypto framework and the accelerator)
2015-06-29 11:52:35 <leo-unglaub> i have to get some food, see you later
2015-06-29 11:52:42 <ncopa> bon apetit
2015-06-29 11:56:13 <kaniini> anyway, padlock is a serious difference between an obsolete server still being able to be useful or not
2015-06-29 11:58:29 <ncopa> i heard an interesting idea about the reason why via/padlock is not more popular on the market: because a big govenrment organisation probably has no backdoors in it
2015-06-29 11:59:05 <ncopa> so this organistaion make it difficult for them and try force people over to intel etc...
2015-06-29 12:03:19 <kaniini> of course there's so many big organisations to choose from
2015-06-29 12:03:29 <kaniini> (who am i kidding it is probably either NSA or GCHQ ;))
2015-06-29 12:07:09 <kaniini> maybe i should call said openssl fork, "angryssl"
2015-06-29 12:07:17 <kaniini> because reading the source code should make you angry due to how bad it is
2015-06-29 12:08:09 <ncopa> :D
2015-06-29 12:09:31 <crow> ncopa regarding that oscam segfault it happens hier if i have oscam started, and i start it one more time (lets say i write start insteed stop)
2015-06-29 12:11:59 <ncopa> ok
2015-06-29 12:21:11 <crow> i have backtrace if needed, i am not sure if init.d or oscam related
2015-06-29 12:22:44 <crow> ncopa http://sprunge.us/ddJI
2015-06-29 12:46:16 <ncopa> crow: do you think you could pass that to oscam devs?
2015-06-29 12:47:46 <clandmeter> http://www.streamboard.tv/oscam/report
2015-06-29 12:48:48 <ncopa> crow: is it with alpien edge?
2015-06-29 13:19:41 <leo-unglaub> ncopa: pidgin update worked
2015-06-29 13:19:43 <leo-unglaub> :)
2015-06-29 13:20:09 <ncopa> good! thanks!
2015-06-29 13:20:31 <kaniini> ...you use pidgin as your irc client???
2015-06-29 13:20:38 <kaniini> why would you do such a terrible thing
2015-06-29 13:20:52 <leo-unglaub> kaniini: because hexchat is worse in some cases
2015-06-29 13:21:09 <leo-unglaub> otr in hexchat is horrible
2015-06-29 13:21:17 <leo-unglaub> and i have to know, i wrote the plugin *g*
2015-06-29 13:21:49 <kaniini> so, let me get this straight.  you're concerned about security yet you're running an irc client with a ton of crash bugs?  :p
2015-06-29 13:22:47 <leo-unglaub> someone once called pidgin a series of working buffer overflows ...
2015-06-29 13:22:51 <leo-unglaub> and this person was right
2015-06-29 13:23:22 <leo-unglaub> on i3wm i use weechat
2015-06-29 13:23:32 <leo-unglaub> but i have not found a nice gtk chat client
2015-06-29 13:23:38 <kaniini> so i took a shower earlier and was thinking more about our openssl problem
2015-06-29 13:24:11 <kaniini> and i think really the way to win here is to build up from the ground up with known good code instead of try to fix it
2015-06-29 13:24:30 <leo-unglaub> kaniini: i could not agree more!
2015-06-29 13:24:47 <leo-unglaub> thats why me and my company use mbedtls (former polarssl) for every project
2015-06-29 13:24:57 <leo-unglaub> however, try telling the apache guys to change there ssl lib
2015-06-29 13:25:03 <leo-unglaub> they ban you for trolling
2015-06-29 13:25:14 <leo-unglaub> so until this is fixed it needs a secure drop in replacement
2015-06-29 13:25:30 <kaniini> right, i was thinking something like a compatibility layer ontop of mbedtls or some other similarly licensed software
2015-06-29 13:25:31 <leo-unglaub> but you are 100% correct if you say security needs to be redone right fromthe ground
2015-06-29 13:25:43 <kaniini> if that makes sense
2015-06-29 13:25:57 <kaniini> something where we know it's good, we just build the openssl api ontop
2015-06-29 13:26:13 <kaniini> i think its doable.  gnutls had a compatibility layer for a while.
2015-06-29 13:26:16 <kaniini> but it was GPL. :(
2015-06-29 13:26:20 <leo-unglaub> i know the polarssl guy, and we have talked alot about an polarssl openssl wrapper
2015-06-29 13:26:36 <kaniini> which obviously is Not Good
2015-06-29 13:26:46 <leo-unglaub> the problem ... its impossible to get right
2015-06-29 13:27:01 <kaniini> it doesn't need to be 100% right
2015-06-29 13:27:18 <leo-unglaub> you are crazy about that  version number think, try creating a drop in replacent that mirrors openssl 100% and is still polarssl
2015-06-29 13:27:19 <kaniini> perfection is the enemy of the good enough
2015-06-29 13:27:21 <leo-unglaub> impossible to do
2015-06-29 13:27:57 <kaniini> leo-unglaub: i mean as a mid-term solution
2015-06-29 13:28:01 <leo-unglaub> ah, i see
2015-06-29 13:28:12 <kaniini> something to port to that is close enough
2015-06-29 13:28:25 <kaniini> that you don't lose motivation
2015-06-29 13:28:45 <kaniini> and then for short-term we just keep openssl
2015-06-29 13:28:55 <kaniini> apache has other ssl modules
2015-06-29 13:29:00 <kaniini> there is mod_gnutls, for example
2015-06-29 13:29:17 <leo-unglaub> well, gnutls is not better
2015-06-29 13:29:18 <kaniini> so you do not really have to use openssl there
2015-06-29 13:29:20 <leo-unglaub> thats the problem
2015-06-29 13:29:27 <kaniini> yes, i am just saying
2015-06-29 13:29:33 <leo-unglaub> yeah, i get your point
2015-06-29 13:29:36 <leo-unglaub> and i agree wit it
2015-06-29 13:29:37 <kaniini> you could make an apache module which uses mbedtls
2015-06-29 13:29:48 <kaniini> or perhaps openssl-portability-layer-to-mbedtls
2015-06-29 13:30:14 <kaniini> it is possible
2015-06-29 13:30:15 <leo-unglaub> if you mange to write such a layer you have a fix job at ARM currently
2015-06-29 13:30:27 <kaniini> because, fedora is openssl-free
2015-06-29 13:30:33 <kaniini> although they used NSS ;)
2015-06-29 13:31:02 <leo-unglaub> nss has a horrible api, however the actual code is not that bad
2015-06-29 13:31:15 <kaniini> NSS's api is basically the same as OpenSSL
2015-06-29 13:31:22 <kaniini> or, rather
2015-06-29 13:31:27 <kaniini> OpenSSL's api is modeled after NSS
2015-06-29 13:31:33 <kaniini> so it is quite similar
2015-06-29 13:31:46 <leo-unglaub> as i sayed: horrible ...
2015-06-29 13:32:11 <kaniini> however, i do not think a portability layer really needs to support all use-cases of openssl
2015-06-29 13:32:16 <kaniini> just 90%
2015-06-29 13:32:33 <kaniini> the idea is you provide enough to make bringup on say, mbedtls, easy enough to make feasible
2015-06-29 13:33:00 <kaniini> and if such a package existed
2015-06-29 13:33:14 <leo-unglaub> i would be okay with 30% ... i mean .. what do people do with openssl really? in 90% of all cases they open a https connection
2015-06-29 13:33:26 <kaniini> we would probably do the work to use it to get rid of openssl
2015-06-29 13:33:34 <kaniini> leo-unglaub: right i am just talking 90% of what people actually do with it
2015-06-29 13:34:23 <kaniini> so basic apps would just build and work against it
2015-06-29 13:34:35 <kaniini> then if you needed something more complex, you have to break out to mbedtls api itself
2015-06-29 13:34:47 <kaniini> and then we provide some wrapper functions to go from an SSL_CTX to an mbedtls ssl_context
2015-06-29 13:35:54 <kaniini> this is what we did with pkgconf
2015-06-29 13:36:07 <kaniini> we simulated the basic usecases and fixed the broken stuff
2015-06-29 13:36:07 <kaniini> and people pretty much seem to be cool with it
2015-06-29 13:37:44 <kaniini> but we started from 0 and started adding in things actually used until we got to having it working for building the entire archive
2015-06-29 13:39:45 <kaniini> leo-unglaub: the problem with mbedtls is it's GPL
2015-06-29 13:40:21 <kaniini> and while it has: https://tls.mbed.org/foss-license-exception
2015-06-29 13:41:22 <leo-unglaub> well, for me personally the GPL was never a problem
2015-06-29 13:43:32 <kaniini> the problem is that as a distribution,
2015-06-29 13:43:44 <kaniini> we cannot necessarily distribute GPL-incompatible + GPL binaries
2015-06-29 13:44:20 <ncopa> seems like firefox segfault on youtube videos now :-/
2015-06-29 13:44:26 <kaniini> leo-unglaub: but i heard mbedtls is switching to Apache License now
2015-06-29 13:44:45 <kaniini> so that could work
2015-06-29 13:46:10 <kaniini> i.e.: Perhaps the biggest change to mbed TLS will hit a bit later this year and has quite a big impact. mbed TLS will change its main license from the GPL to the liberal Apache license. This is great news for a lot of open source projects and commercial projects that want to use mbed TLS, but could not use GPL licensed code or afford a commercial license. The Apache license allows even easier use of
2015-06-29 13:46:10 <kaniini> mbed TLS in open source projects and no more commercial license is needed to use PolarSSL in your closed source (commercial) projects.
2015-06-29 13:47:15 <mitchty> @work gpl is basically acid, bsd3/mit is preferred, haven't had to hit apache yet so no idea there, but lawyers are not fond of any gpl
2015-06-29 13:47:32 <mitchty> then again we ship binaries so there is that
2015-06-29 13:48:45 <kaniini> apache is fine
2015-06-29 13:49:25 <kaniini> it's pretty much the industry standard "do whatever you want but don't sue us for patents" license
2015-06-29 13:49:40 <mitchty> ah, yeah doubt the lawyers would care about that then
2015-06-29 13:50:23 <ScrumpyJack> yeah, same fear of GPL@work here too, and we don't even ship code. Listening to lawyers on GPL is funny. When I'm bored i go poke a lawyer
2015-06-29 13:50:46 <ncopa> :)
2015-06-29 13:50:50 <kaniini> ScrumpyJack: GPL does have real implications if you are shipping code
2015-06-29 13:51:03 <mitchty> the best part about lawyer talk is when they won't email anything and just tell you verbally
2015-06-29 13:51:21 <mitchty> ah, i see, so if things go bad I know i'm going under a bus
2015-06-29 13:51:23 <kaniini> and using GPL for a *library* is *insane*
2015-06-29 13:51:30 <ScrumpyJack> kaniini: I'm not prepared to have that conversation with you right now, you're fired up enough as it is ;)
2015-06-29 13:52:10 <mitchty> well if we ship binaries with gpl even static linked in the lawyers I've talked to have said they interpret that to mean we would have to gpl our source
2015-06-29 13:52:14 <kaniini> ScrumpyJack: what conversations are there to have?  GPL does infect other modules, it is designed to
2015-06-29 13:53:27 <kaniini> 
ACTION is not anti-GPL, just anti-GPL in core system libraries - if you want to use GPL for your app do whatever you want

2015-06-29 13:53:47 <ScrumpyJack> sorry, i just read what i wrote. I mean't we don't ship binaries (or anything)
2015-06-29 13:54:21 <kaniini> ScrumpyJack: as a user of a GPL program it doesn't matter, your lawyers are nuts
2015-06-29 13:54:26 <mitchty> ScrumpyJack it normally doesn't matter until you need to ship things to customers, or matters less unless agpl/gpl3
2015-06-29 13:54:32 <kaniini> it only matters if you are shipping
2015-06-29 13:54:52 <kaniini> well AGPL is a nightmare in and of itself, but nobody uses that
2015-06-29 13:54:57 <mitchty> internally as users we can use anything
2015-06-29 13:55:41 <ncopa> i wonder what broke firefox
2015-06-29 13:56:11 <kaniini> leo-unglaub: so i think what i want to do about openssl is replace it with mbedtls and provide a shim library to make porting apps over easy (and then, you know, upstream the patches and perhaps make a site like "we can make openssl die today" or such)
2015-06-29 13:56:36 <kaniini> that way we solve the real problem
2015-06-29 13:56:44 <kaniini> libressl is just like a bandaid
2015-06-29 13:56:59 <leo-unglaub> if you want to be really cool you have to name the site: "hasopenssldiesyet" ;(
2015-06-29 13:57:00 <leo-unglaub> ;)
2015-06-29 13:57:08 <leo-unglaub> the yet is important for it to be hip *g*
2015-06-29 13:58:32 <ScrumpyJack> We ship stuff, but I really don't care. Our lawyers are very concerned, but mainly out of ignorance.
2015-06-29 14:00:23 <kaniini> well, it is best to have an IP lawyer review it not just any lawyer
2015-06-29 14:01:58 <mitchty> as the one lawyer told me, we interpret things in the worst case scenario, the only true test is what a judge will decide
2015-06-29 14:10:28 <bCQc0P3DIqbh> what about nacl? since we're now talking fixing things, irrespective of porting efforts
2015-06-29 14:41:53 <fcolista> ncopa, i'm ready to push v.1.6.3 v4l-utils
2015-06-29 14:42:08 <fcolista> how can I check if ABI compat breaks?
2015-06-29 14:42:08 <ncopa> ok?
2015-06-29 14:42:12 <ncopa> checkapk
2015-06-29 14:42:14 <fcolista> ah
2015-06-29 14:42:26 <ncopa> with the new tar it gives lots of warnings
2015-06-29 14:42:26 <fcolista> i was looking for abuild --help and apk --help :)
2015-06-29 14:42:36 <ncopa> so i do: checkapk 2>/dev/null
2015-06-29 14:44:51 <fcolista> umh
2015-06-29 14:44:57 <fcolista> http://sprunge.us/PaJB
2015-06-29 14:45:05 <ncopa> should be safe
2015-06-29 14:45:17 <fcolista> it's just add a file
2015-06-29 14:45:21 <ncopa> yes
2015-06-29 14:45:51 <fcolista> ok let's see :)
2015-06-29 14:46:07 <fcolista> anyway, pkgs with v4l-utils are few
2015-06-29 17:37:19 <ScrumpyJack> People are /query and/or emailing saying that they've been using this or that aport and it works fine. I'd like to be able to ask them to pop that into a feedback form for the package, as evidence of testing. Is there such a thing already setup?
2015-06-29 17:38:29 <ScrumpyJack> It would work for package requests in b.a.o, they could just at to the bug number page, but not all packages have reqeusts, obvs.
2015-06-29 17:40:10 <ScrumpyJack> s/at/add
2015-06-29 17:43:50 <barthalion> nope
2015-06-29 17:44:05 <barthalion> I just got used to getting e-mails or messages on IRC
2015-06-29 19:49:33 <BitL0G1c> can the version of dialog in edge be moved to main please ? it's needed for cryptsetup benchmark to work in my partition script https://asciinema.org/a/22619 --prgbox is broken in dialog from 2014
2015-06-29 19:54:07 <vkris> ahills: http://imgur.com/a/COUNt/all ;), still not complete, but functional
2015-06-29 19:54:44 <vkris> unable to find a local [AL]uminium fabricator
2015-06-29 19:55:17 <vkris> and getting small fan, small motherboard... seems like huge problem
2015-06-29 19:55:29 <ahills> looks pretty solid
2015-06-29 19:55:38 <ahills> is that for outdoor/industrial use or just home?
2015-06-29 19:55:54 <vkris> but hold-on, if we can send these things to Mars, then possible ;)
2015-06-29 19:56:35 <vkris> just home, plan to put all my Internet-of-things on it + rss aggregator
2015-06-29 19:56:52 <vkris> plus home share portal
2015-06-29 19:57:48 <vkris> getting 170x170mm board in India seems like impossible
2015-06-29 20:04:54 <vkris> if somebody finances, I would rather buy those power tools to make the [AL]uminium box
2015-06-29 20:05:17 <vkris> any maybe show to local vendors and get some attentions ;)
2015-06-30 06:29:18 <ncopa> morning
2015-06-30 06:30:03 <ncopa> clandmeter: this is a problem: http://git.alpinelinux.org/cgit/aports/commit/main/freetype?id=0ecaf695611d0af6f4c6610d5c1264de9905e19e
2015-06-30 06:30:18 <ncopa> freetype needs harfbuzz and harfbuzz needs freetype
2015-06-30 06:30:22 <ncopa> circular dep
2015-06-30 06:41:49 <clandmeter> ncopa: ok
2015-06-30 06:44:15 <clandmeter> what can we do?
2015-06-30 06:44:45 <clandmeter> ok i see you already comitted it
2015-06-30 06:46:07 <ncopa> i had to revert harfbuzz support in freetype
2015-06-30 06:46:13 <ncopa> we should ask upstream
2015-06-30 06:49:01 <barthalion> well, upstream introduced it
2015-06-30 06:54:54 <ncopa> which is why I want to ask them
2015-06-30 06:55:15 <ncopa> but that seems to be the norm nowdays, circular deps and bootstrapping
2015-06-30 06:55:19 <ncopa> pkg-config started it
2015-06-30 06:55:27 <ncopa> openjdk
2015-06-30 06:55:35 <ncopa> now freetype/harfbuzz
2015-06-30 07:18:10 <ScrumpyJack> morning
2015-06-30 07:21:05 <ScrumpyJack> bluefish and mate-control-center packages clash. Bluefish trying to overwrite ... owned by mate-control-center
2015-06-30 07:21:10 <ScrumpyJack> how do i fix that?
2015-06-30 07:50:15 <ncopa> what file is it?
2015-06-30 07:50:27 <ncopa> one of the packaes might need rename the file
2015-06-30 08:20:01 <ncopa> \o/
2015-06-30 08:20:06 <ncopa> firefox youtube works again
2015-06-30 13:42:29 <vkris> ncopa: does zeromq v4.1.2 require patching or I need to update/upgrade ?
2015-06-30 13:42:54 <ncopa> ?
2015-06-30 13:43:20 <vkris> zeromq v4.1.2 <-- new release, bump pls
2015-06-30 13:43:31 <vkris> to see what happens
2015-06-30 13:49:17 <vkris> ncopa: there is an older version of mqtt.js available for using on http://build.alpinelinux.org/
2015-06-30 13:49:28 <vkris> this would not give errors at bottom of page
2015-06-30 13:49:47 <ncopa> what error do you see?
2015-06-30 13:50:19 <vkris> http://sprunge.us/LMjb
2015-06-30 13:51:08 <vkris> but I am testing on iceweasel 21.0
2015-06-30 13:52:52 <vkris> seems ok in chromium 27.0.1453.110 Debian 7.1 (202711)
2015-06-30 13:53:01 <barthalion> dude, 21?
2015-06-30 13:53:09 <barthalion> to say that you are obsolete is not enough
2015-06-30 13:53:15 <barthalion> 38 is out
2015-06-30 13:53:39 <vkris> barthalion: I need for webapp deveopment
2015-06-30 13:54:07 <vkris> atleast 1.5yrs old browsers
2015-06-30 13:54:31 <barthalion> I doubt anyone is still on 21
2015-06-30 13:54:44 <barthalion> firefox auto-updates on windows and osx
2015-06-30 13:57:10 <vkris> iirc, paho mqtt was reported to be buggy on #mqtt
2015-06-30 13:58:27 <ncopa> sigh
2015-06-30 13:58:36 <ncopa> xen does not build with gcc5
2015-06-30 13:59:05 <ncopa> it needs patches for the firmware, seabios
2015-06-30 13:59:14 <ncopa> which is pulled from git at build time
2015-06-30 13:59:19 <ncopa> how do i patch it?
2015-06-30 13:59:45 <ncopa> i suppose i could do what fedora does and build seabios as separate package
2015-06-30 14:00:20 <vkris> pulled into src dir ?
2015-06-30 14:00:26 <ncopa> yes
2015-06-30 14:00:47 <ncopa> so the needed patch does not apply because the file it patches it not yet there
2015-06-30 14:02:15 <vkris> maybe prep the full src before
2015-06-30 14:13:40 <ncopa> fedora ships a separate seabios package
2015-06-30 14:13:48 <ncopa> and tell xhen to use system seabios
2015-06-30 14:13:53 <ncopa> i think thats a better approach
2015-06-30 14:22:13 <vkris> so qemu can also use it ?
2015-06-30 14:22:25 <ncopa> yes, i believe so
2015-06-30 16:24:49 <alacerda> any idea of why this is happening: http://hastebin.com/xuqiyibaru.vbs ?
2015-06-30 16:26:57 <clandmeter> alacerda: did you apk upgrade?
2015-06-30 16:31:26 <alacerda> i did
2015-06-30 16:35:03 <alacerda> clandmeter, ¬¬' thanks I issued upgrade againg and there was some other packages resting to be upgraded
2015-06-30 16:35:09 <alacerda> now it is working
2015-06-30 16:38:09 <clandmeter> ;-)
2015-06-30 16:38:40 <clandmeter> im very close in getting chromium build :)
2015-06-30 16:40:34 <AmatCoder> <clandmeter:
2015-06-30 16:40:46 <AmatCoder> clandmeter: nice!
2015-06-30 16:58:30 <clandmeter> i should really add ccache setup to my profile.
2015-06-30 17:00:59 <clandmeter> [16774/16774] LINK chrome
2015-06-30 18:59:48 <kaniini> clandmeter: does it run :D
2015-06-30 19:01:43 <vkris> kaniini: have you used low-end gigabyte mb`s ?
2015-06-30 19:01:59 <kaniini> vkris: no
2015-06-30 19:02:07 <kaniini> vkris: for server deployments i use supermicro pretty much
2015-06-30 19:02:40 <kaniini> 
ACTION gave up on the whole "linux as desktop" concept some time ago

2015-06-30 19:03:50 <vkris> Initially I tried to get supermicro, but no vendors support in locally
2015-06-30 19:04:02 <vkris> I even got quotes and was almost ready for it
2015-06-30 19:04:39 <scadu> kaniini: so what do you use on desktop?
2015-06-30 19:04:47 <vkris> only few cities here have branch, but marketting and billings are wierd
2015-06-30 19:05:11 <vkris> scadu: good question
2015-06-30 19:05:13 <vkris> ;)
2015-06-30 19:06:10 <kaniini> well few years ago my employer standardized on mac, so i just use that now
2015-06-30 19:06:12 <vkris> I am guessing he is so busy with admin stuffs , only uses elinks for browsing
2015-06-30 19:06:37 <kaniini> although mac is getting really broken these days
2015-06-30 19:06:40 <kaniini> so i might revisit the idea
2015-06-30 19:06:44 <kaniini> dunno.
2015-06-30 19:06:54 <scadu> vkris: sth went wrong ;f
2015-06-30 19:06:56 <vkris> mac is a taboo, except for few cities here
2015-06-30 19:11:08 <vkris> something makes me feel desktop market is dead here
2015-06-30 19:11:56 <vkris> and iirc Intel even stopped desktop mb manufacturing
2015-06-30 19:12:03 <vkris> only server now
2015-06-30 19:12:41 <vkris> and I was just getting started ;-))
2015-06-30 21:02:06 <clandmeter> kaniini: not yet, i have no idea how to add libs/ldflags this ninja crap
2015-06-30 21:02:28 <kaniini> :D
2015-06-30 21:03:14 <clandmeter> so i have to remove all refs to backtrace, which is not difficult, but time consuming.
2015-06-30 21:42:34 <clandmeter> kaniini: it did compile and link now :D now to cleanup this mess.
2015-06-30 22:30:43 <clandmeter> chromium-43.0.2357.130-r0.apk
2015-06-30 22:30:47 <clandmeter> \o/
2015-06-30 22:48:23 <crow> something new again http://blogs.aws.amazon.com/security/post/TxCKZM94ST1S6Y/Introducing-s2n-a-New-Open-Source-TLS-Implementation
2015-06-30 22:52:35 <pnutzh4x0r> clandmeter: how well does the chromium aport work?
2015-06-30 22:53:00 <clandmeter> i have no idea
2015-06-30 22:53:21 <pnutzh4x0r> oh, but it builds?
2015-06-30 22:53:29 <clandmeter> yes, its the first step
2015-06-30 22:53:42 <clandmeter> next is to make it build on grsec
2015-06-30 22:53:48 <pnutzh4x0r> ok, cool
2015-06-30 22:53:51 <clandmeter> then to fix ugly patches
2015-06-30 22:53:56 <clandmeter> submit them upsream
2015-06-30 22:54:04 <clandmeter> and probably fix sandboxing
2015-06-30 22:54:59 <clandmeter> its wip, but beeing able to build it is a good step in the right direction.
2015-06-30 22:55:21 <clandmeter> damn mosquittos
2015-06-30 22:55:26 <clandmeter> im going to bed.