2022-12-01 10:30:24 <chereskata> Wouldn't it be nice to delete invalid flags from https://pkgs.alpinelinux.org?
2022-12-01 10:30:59 <psykose> probably
2022-12-01 10:31:29 <chereskata> I found some, should i post them here?
2022-12-01 10:31:48 <psykose> i already know them all probably
2022-12-01 10:31:54 <psykose> just no access to do that
2022-12-01 10:31:57 <psykose> ikke: hey actually..
2022-12-01 10:32:10 <chereskata> Do you know who is the maintainer of pkgs.alpinelinux.org?
2022-12-01 10:35:49 <ikke> psykose: hi, nice to see you here again 👋
2022-12-01 10:36:05 <psykose> ikke: https://img.ayaya.dev/vHxHLlUucqvQ yeah as you can see.. something happened..
2022-12-01 10:36:38 <psykose> do you mind giving me access to whatever to remove false flags from pkgs.a.o
2022-12-01 10:38:18 <chereskata> is there any interest to bring the torbrowser to alpine?
2022-12-01 10:38:48 <psykose> the dns resolver library thing they have does not work with musl
2022-12-01 10:38:52 <psykose> aside from that it's just firefox
2022-12-01 10:39:21 <chereskata> helping to obfuscate traffic by using it for everyday tasks would be great   
2022-12-01 10:40:02 <ptrc> it gets tiring after a while
2022-12-01 10:40:14 <psykose> what does
2022-12-01 10:40:33 <psykose> ah, tor browser
2022-12-01 10:40:39 <psykose> yeah
2022-12-01 10:40:48 <chereskata> I use it everyday with tails or on mobile
2022-12-01 10:40:53 <psykose> but - it means once you stop you get to experience the s p e e d of networking, for sure
2022-12-01 10:41:00 <chereskata> as im a nojs surfer i dont care
2022-12-01 10:41:29 <chereskata> having half a gigabit link does help btw
2022-12-01 10:41:50 <psykose> it doesn't, as the former owner of symmetric gigabit
2022-12-01 10:41:57 <psykose> i expect websites to load in <1s, not 35
2022-12-01 10:42:26 <chereskata> i managed to speed it up by running a local recursive resolver
2022-12-01 10:42:31 <chereskata> on my router
2022-12-01 10:44:35 <psykose> regardless, if it actually worked i would probably package it
2022-12-01 10:44:48 <psykose> i don't remember the specific dns issues anymore
2022-12-01 10:45:09 <psykose> something about this https://gitlab.torproject.org/tpo/core/tor/-/issues/40248
2022-12-01 11:26:20 <chereskata> I found some discussions in these months and years of the musl mailing list
2022-12-01 11:29:22 <chereskata> 2012-05 2014-06 2016-05 2016-06 2017-01 2017-10 2018-06 2021-01 2022-05 2022-06 2022-09
2022-12-01 11:29:36 <chereskata> https://www.openwall.com/lists/musl/
2022-12-01 11:29:52 <chereskata> p.s. some discussions with dns as topic
2022-12-01 11:31:14 <chereskata> since i have not found a sidewide search, i just scraped every month and grepped over the html files
2022-12-01 12:06:46 <chereskata> nano-syntax is not dependend on nano
2022-12-01 12:07:07 <chereskata> i can remove nano without nano-syntax beeing removed
2022-12-01 12:21:44 <chereskata> ty for the fast fix :)
2022-12-01 12:39:12 <raspbeguy> hello, as I indicated in !41870 and !41871 I wish those upgrade to be backported into stable version. What is the procedure?
2022-12-01 12:52:41 <psykose> you make a merge request against 3.17-stable
2022-12-01 12:52:48 <psykose> and make your branch from it too
2022-12-01 13:30:05 <chereskata> i have noticed that the dbus-* cookie is world writable
2022-12-01 13:30:22 <chereskata> on debian it was locked
2022-12-01 13:37:55 <psykose> which one
2022-12-01 13:38:02 <chereskata> oh sorry
2022-12-01 13:38:07 <chereskata> the one in /tmp
2022-12-01 13:38:13 <chereskata> let me check the others too
2022-12-01 13:38:46 <psykose> what does that let you do
2022-12-01 13:39:24 <chereskata> removing it by accident, when doing rm-rf /tmp/* and breaking the dbus for the session 
2022-12-01 13:39:33 <chereskata> :D
2022-12-01 13:40:32 <chereskata> other ones seem ok 
2022-12-01 16:17:49 <ncopa> does anybody use buildlab that comes with abuild?
2022-12-01 16:19:11 <ncopa> i'd like to remove it
2022-12-01 16:19:18 <ikke> never used it
2022-12-01 16:21:18 <psykose> forgot it even existed
2022-12-01 16:25:43 <ncopa> I have never used it myself
2022-12-01 16:25:51 <ncopa> and now we have abuild rootbld
2022-12-01 20:17:06 <EvTheFuture> ncopa: Just checking if you've had any time to take alook if you could recreate the problem described in issue 14364?
2022-12-01 20:43:51 <Newbyte> Is there any modprobe in Alpine that can handle zstd-compressed kernel modules?
2022-12-01 20:44:19 <Newbyte> modprobe: can't load module ext4 (kernel/fs/ext4/ext4.ko.zst): invalid module format
2022-12-01 20:44:20 <Newbyte> getting this
2022-12-01 20:44:26 <Newbyte> (I'm not using Alpine's mkinitfs)
2022-12-01 20:44:42 <ikke> Newbyte: try kmod
2022-12-01 20:44:50 <Newbyte> Thanks
2022-12-01 20:46:01 <ikke> https://pkgs.alpinelinux.org/contents?file=modprobe&path=&name=&branch=edge&arch=x86_64
2022-12-02 02:53:41 <psykose> algitbot: ping
2022-12-02 02:55:34 <minimal> Avon calling?
2022-12-02 03:01:20 <psykose> is that what yer da sells
2022-12-02 03:02:32 <minimal> "Vila, teleport now!"
2022-12-02 03:03:07 <minimal> (UK 70s/80s SciFi reference)
2022-12-02 03:04:27 <psykose> :)
2022-12-02 07:27:21 <ncopa> EvTheFuture: I was not able to reproduce. 3d worked in my vm. I wonder if it can be related wayland?
2022-12-02 08:47:40 <EvTheFuture> ncopa: Maybe or the fact I use amdgpu?
2022-12-02 08:48:00 <EvTheFuture> ncopa: Is it possible for you to try Wayland?
2022-12-02 08:52:37 <ncopa> Maybe later today
2022-12-02 08:53:33 <ncopa> i think there was a qemu- package that I had installed that you didn’t 
2022-12-02 08:55:58 <ncopa> qemu-vhost-user-gpu IIRC
2022-12-02 08:56:11 <ncopa> not sure it makes any diff
2022-12-02 14:30:19 <ncopa> ptrc: re https://gitlab.alpinelinux.org/alpine/abuild/-/issues/10080
2022-12-02 14:30:54 <ncopa> I asked https://chat.openai.com/chat "Create a regexp that matches an rcf822-compliant mailbox"
2022-12-02 14:31:08 <ncopa> here is what I got: ^(?:"[^"]+"|[^<>@\s,]+)(?:@[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*)$
2022-12-02 14:31:31 <ncopa> pretty impressive :)
2022-12-02 14:53:46 <pj[m]> ncopa: did you verify that it's actually RFC822 compliant though?
2022-12-02 15:31:15 <mercenary> http://www.ex-parrot.com/~pdw/Mail-RFC822-Address.html
2022-12-02 15:37:55 <ncopa> pj[m]: close enough
2022-12-02 15:38:20 <ncopa> i wonder if we should accept "
2022-12-02 15:38:24 <ncopa> >>> ERROR: ossp-uuid: '"Andrzej Trzaska <atrzaska2@gmail.com>"' is not a valid rfc822 address
2022-12-02 15:38:47 <pj[m]> :)
2022-12-02 15:38:59 <ncopa> # Maintainer: "Andrzej Trzaska <atrzaska2@gmail.com>"
2022-12-02 15:39:12 <ncopa> I guess that's valid rfc822
2022-12-02 15:39:21 <ncopa> but should we accept it=
2022-12-02 15:39:22 <ncopa> ?
2022-12-02 15:39:37 <ncopa> I also disallow leading and trailing spaces
2022-12-02 15:40:10 <ncopa> IMHO, the stricter the better
2022-12-02 15:58:55 <ptrc> ncopa: but reasonably strict; for now we don't even allow some valid given names
2022-12-02 15:59:39 <ptrc> having the whole name in quotes should be valid too
2022-12-02 16:00:32 <ptrc> oh, just saw the commit
2022-12-02 16:00:39 <ptrc> sounds good
2022-12-02 19:02:59 <raspbeguy> What could be the reason to keep a port in testing instead of moving it to community or main ?
2022-12-02 19:04:19 <raspbeguy> I found a wiki page about installing kubernetes on alpine that says to enable testing repo to install kubelet/kubeadm
2022-12-02 19:15:01 <ikke> raspbeguy: lack of feedback from users, dependencies in testing, other issues with the package that prevent it from being included in stable releases, or the maintainer just forgot about it
2022-12-03 08:50:52 <EvTheFuture> ncopa: I will try this
2022-12-03 08:51:22 <EvTheFuture> ncopa: I will try with qemu-vhost-user-gpu installed, thank you.
2022-12-03 09:43:22 <vkrishn> newsboat seems ok in v3.16(aarch64), but has date problem in armv7(v3.15, v│3.16)
2022-12-03 09:44:26 <vkrishn> would appreciate if someone could test and confirm, I did this on a non-standard alpinelinux install
2022-12-03 10:00:08 <EvTheFuture> ncopa: unfortunately installing the qemu-vhost-user-gpu package in the host didn't help. still no acceleration inside the guest running ubuntu 20.04
2022-12-03 10:05:33 <psykose> what date problem specifically
2022-12-03 10:16:32 <psykose> ah, i can reproduce it
2022-12-03 10:16:33 <psykose> blech
2022-12-03 10:19:08 <vkrishn> thanks, can it get fixed from al side, or needs poking upstream ?
2022-12-03 10:21:11 <vkrishn> I was trying to release an update for https://github.com/insteps/nbreader on, but got stuck
2022-12-03 10:22:06 <vkrishn> for php7.4+, php8.0+
2022-12-03 10:27:47 <liske> I'm trying to get a backtrace for #14408 - but did not succeed. I've build the package locally with !strip and run it in foreground with ulimit -c unlimited... but no core file is created. What am I doing wrong?
2022-12-03 10:28:40 <psykose> did you set your kernel.core_pattern to put corefiles somewhere else than cwd
2022-12-03 10:30:42 <liske> oh it might be a permission issue... bird runs as user bird and I've started it from /root
2022-12-03 10:31:09 <psykose> that would also make sense
2022-12-03 10:32:02 <liske> the pattern was default (kernel.core_pattern = core), changed it to put it in /tmp
2022-12-03 10:33:48 <liske> (and yes, /proc/$(pidof bird)/cwd points to /root)
2022-12-03 10:34:11 <liske> psykose: thanks!
2022-12-03 10:34:40 <psykose> been bitten by the unwritable cwd many times :)
2022-12-03 10:38:03 <skarnet> it just gave you the bird
2022-12-03 11:03:08 <vkrishn> msgs from algibot don't make to irclogs, but its a valid irc msg
2022-12-03 11:04:07 <ikke> algitbot is what's logging the messages, and it does not record what it sends itself. 
2022-12-03 11:05:04 <psykose> ikke: did you see my request about getting access to the pkgs.a.o flagged thing so i can remove all the incorrect/stale flags the other day
2022-12-03 11:09:47 <liske> psykose: bird crashed again but no core file has been created
2022-12-03 11:10:27 <psykose> hmm
2022-12-03 11:10:45 <psykose> well if the core pattern is to a writable folder, and it's sigabrt
2022-12-03 11:10:56 <psykose> as long as it doesn't change what it does with sigaction in the program it should make one
2022-12-03 11:11:09 <psykose> if they call sigaction for sigabrt to do something else it wouldnt
2022-12-03 11:11:12 <psykose> afaik
2022-12-03 11:11:26 <psykose> `ulimit -c unlimited` also has to be in the same shell
2022-12-03 11:11:43 <psykose> i am not sure what happens if they fork out of it or something, but foreground should be fine
2022-12-03 11:12:55 <liske> grep core /proc/$(pidof bird)/limits shows unlimited/unlimited
2022-12-03 11:12:56 <vkrishn> dont't make the ai's angry ;)
2022-12-03 11:13:18 <psykose> that should be ok then
2022-12-03 11:13:29 <psykose> what is the core_pattern exactly
2022-12-03 11:13:37 <liske> and there is only a single pid
2022-12-03 11:13:54 <liske>  /tmp/core.%p
2022-12-03 11:14:04 <psykose> hmm
2022-12-03 11:14:07 <liske> is it a problem that /tmp is world writable?
2022-12-03 11:14:13 <psykose> no, that should work
2022-12-03 11:14:21 <psykose> i use /tmp/core-%e-%s-%u-%g-%p-%t
2022-12-03 11:14:24 <psykose> tmp world writable
2022-12-03 11:19:34 <liske> it doesn't look like the change sigabrt in the sources: https://gitlab.nic.cz/labs/bird/-/blob/2.0.10/sysdep/unix/main.c#L652
2022-12-03 11:20:13 <psykose> indeed not
2022-12-03 11:20:51 <psykose> i can also make it drop a core myself with kill -ABRT
2022-12-03 11:20:58 <psykose> just bird -f and that
2022-12-03 11:23:10 <liske> oh
2022-12-03 11:23:22 <liske> but it won't work with -u bird!
2022-12-03 11:23:44 <liske> running as root I got a Aborted (core dumped)
2022-12-03 11:23:59 <psykose> gonna guess as -u bird it reforks into the -u
2022-12-03 11:24:09 <psykose> but it was still unlimited..
2022-12-03 11:25:02 <psykose> yeah, doesn't work for me either with -u
2022-12-03 11:25:21 <liske> for debugging running as root should be OK since bird doesn't keep data on the fs which would mess up file permissions/ownership
2022-12-03 11:25:36 <liske> psykose: thanks, again :)
2022-12-03 11:26:06 <psykose> and if i try and search -u in this gitlab.nic.cz instance it 500's their backend
2022-12-03 11:26:10 <psykose> amazing stuff
2022-12-03 11:30:04 <psykose> ok, figured it out
2022-12-03 11:30:09 <psykose> with -u they drop_uid
2022-12-03 11:30:31 <psykose> drop_uid does set_capabilities(caps) and drops the rest
2022-12-03 11:31:51 <psykose> as far as i can tell from core.5, you don't get coredumps with either a suid program ran as another user or with file caps
2022-12-03 11:32:03 <psykose> unless prctl PR_SET_DUMPABLE
2022-12-03 11:32:50 <liske>  good to know
2022-12-03 11:33:03 <psykose> i've no idea what is really happening here though
2022-12-03 11:33:15 <psykose> look in sysdep/linux/syspriv.h if you want
2022-12-03 11:33:19 <psykose> bottom of the file
2022-12-03 11:33:49 <psykose> the cap list includes cap_net_admin so it's basically godmode with a user anyway, haha
2022-12-03 11:33:56 <psykose> but yeah i assume this breaks cores
2022-12-03 11:33:59 <psykose> let me try a one line patch
2022-12-03 11:36:41 <psykose> ayy, did it
2022-12-03 11:37:04 <psykose> https://img.ayaya.dev/qNm239386RoI
2022-12-03 11:37:12 <psykose> now they work :)
2022-12-03 11:41:24 <liske> I'm still waiting that a change in the DFZ crashes my birds again, sometimes it takes minutes, sometimes it takes hours
2022-12-03 11:41:33 <psykose> yeah, not fun :/
2022-12-03 11:41:45 <psykose> hopefully just the one backtrace will be enough to figure out what the issue is
2022-12-03 11:41:47 <liske> dunno if it happens running as root at all
2022-12-03 11:42:53 <liske> I have a number of bird instances (1.6 debian, 2.0.8 alpine, 2.0.10 alpine) but only two running on 3.17 are crashing
2022-12-03 11:43:24 <liske> other alpine 3.17 hosts are not affected
2022-12-03 11:43:33 <psykose> strange
2022-12-03 11:44:29 <liske> luckily it is a staging environment... first time I use bird with full table in a Linux VRF
2022-12-03 11:47:55 <psykose> you can send that upstream if you want, doesn't sound like a bad feature to drop the cores
2022-12-03 11:49:56 <liske> sounds like a good idea and add patch for 2.0.10 on alpine for know
2022-12-03 11:51:11 <psykose> once upstream acks it, sure
2022-12-03 11:51:15 <psykose> i don't write code ;)
2022-12-03 12:29:58 <ikke> psykose: I did see it, but I need to prepare something first
2022-12-03 12:30:03 <psykose> sure sure
2022-12-03 12:50:03 <Newbyte> Can I ask about apk-tools development questions in here, or is there a better channel for taht?
2022-12-03 13:02:08 <psykose> you can LD_PRELOAD libapk from the build and then run the apk from the build folder to test that
2022-12-03 13:02:30 <psykose> aside from that it's safe to install anyway
2022-12-03 13:02:48 <psykose> (as long as you are on the stable branch of it)
2022-12-03 13:05:49 <ikke> Newbyte: this is as good as it gets
2022-12-03 13:08:38 <Newbyte> Yeah, I was wondering how to best test this: https://gitlab.alpinelinux.org/alpine/apk-tools/-/merge_requests/116
2022-12-03 13:08:51 <Newbyte> But it seems like someone's already figured out my question
2022-12-03 13:08:52 <Newbyte> Thanks
2022-12-03 13:16:21 <Newbyte> So I'd do something like $ LD_PRELOAD src/libapk.so.3.12.0 apk add?
2022-12-03 13:19:45 <psykose> ./src/apk add
2022-12-03 13:20:30 <Newbyte> Oh, makes sense
2022-12-03 13:20:33 <Newbyte> Thanks
2022-12-03 14:00:15 <q66> Newbyte: that PR is just not necessary
2022-12-03 14:00:17 <q66> yes|apk ...
2022-12-03 14:01:13 <q66> also when using meson to build the binary has rpath set when run in-tree, so no preload needed
2022-12-03 14:03:20 <psykose> there is no meson on 2.x and you cannot use a master branch apk on alpine for obvious reasons
2022-12-03 14:03:33 <q66> you can use master branch apk on alpine
2022-12-03 14:03:38 <q66> for most part
2022-12-03 14:03:46 <psykose> did it actually keep the full repo backwards compat or something
2022-12-03 14:03:49 <q66> yes
2022-12-03 14:03:54 <q66> it supports v2 and v3 indexes and packages
2022-12-03 14:04:13 <psykose> that's epic nice
2022-12-03 14:04:17 <psykose> didn't know :)
2022-12-03 14:05:31 <q66> even stuff like abuild should work because the existing commands use v2 stuff and v3 equivalents have new commands
2022-12-03 14:07:35 <psykose> poggers
2022-12-03 16:26:23 <Newbyte> q66: that's a solution I hadn't thought of, thanks. however, it would mean changing every invocation of apk, which would mean more substantial code changes on our end than setting an environment variable would.
2022-12-03 16:27:39 <Newbyte> but if you want to reject my patch that's fine. I will just have to do it the way you suggested
2022-12-03 17:17:18 <ikke> An advantage of the environment variable could be that if the safe answer to a question is no, apk itself can default to that rather than answering yes to everything
2022-12-03 18:03:08 <pj[m]> @Newbyte don't ship /etc/apk/interactive then
2022-12-03 18:03:18 <Newbyte> pj: no
2022-12-03 18:03:29 <Newbyte> unless you have a good reason not to we're not changing that
2022-12-03 18:04:11 <pj[m]> You are trying to fix issue in apk not caused by apk
2022-12-03 18:04:35 <Newbyte> Interactive mode is a feature of apk
2022-12-03 18:04:51 <pj[m]> apk is not interactive by default
2022-12-03 18:05:16 <ikke> Sometimes you want to opt-in to a feature, but then selectively opt-out
2022-12-03 18:05:24 <psykose> if yer gonna make suggestions at least have them make sense
2022-12-03 18:06:43 <Newbyte> pj: interactive apk is useful for postmarketOS as it can be hard to recover a transaction that didn't go the way you thought it would when you're out and about and using a phone since there's no changing tty or sticking in a USB to fix it
2022-12-03 18:07:05 <Newbyte> or whatever else you may do on a desktop
2022-12-03 18:07:28 <Newbyte> with interactive apk you can see that oops I don't actually want to remove that package before apk starts doing it
2022-12-03 18:08:36 <panekj> sounds like PEBSAC
2022-12-03 18:09:00 <psykose> problem exists between stateofbeinginpoland and chair?
2022-12-03 18:09:07 <panekj> screen
2022-12-03 18:09:35 <Newbyte> panekj: you see, you're not always comfortably sitting in a chair when using a phone
2022-12-03 18:09:47 <panekj> sounds like a you issue
2022-12-03 18:09:54 <panekj> I'm always in my chair
2022-12-03 18:10:00 <psykose> i can't tell if this is just trolling at this point
2022-12-03 18:10:57 <panekj> you're an accomplice 
2022-12-03 18:11:12 <psykose> do you need me to be your mom and tell you off formally or
2022-12-03 18:12:29 <panekj> if it makes you feel better, go off, I'm formally telling you and others that this whole "interactive" situation is stupid and not a problem to be fixed in apk
2022-12-03 18:13:04 <Newbyte> panekj: I suggest you give my MR a big fat downvote
2022-12-03 18:13:28 <psykose> there is nothing to "fix", it's a toggle
2022-12-03 18:13:37 <psykose> same way -i is there in the first place as well as the file checking existence
2022-12-03 18:14:42 <psykose> i would say we should probably delete add/del too, since one can edit the world file by hand, more options are pointless
2022-12-03 18:15:39 <panekj> we should go back to the days when apk was made in shell scripts
2022-12-03 18:16:26 <Newbyte> panekj: it's one sed on /etc/apk/repositories away
2022-12-03 18:16:51 <Newbyte> unless, of course, those old mirrors aren't up any more
2022-12-03 18:16:53 <Newbyte> that would be sad
2022-12-03 18:17:15 <psykose> probably in history somewhere and not as a mirror but i forget
2022-12-03 18:17:16 <panekj> can't wait for ALPINE_FRONTEND=noninteractive
2022-12-03 18:17:27 <psykose> ah yes the classic
2022-12-03 18:17:41 <psykose> you add one feature to your program and tomorrow you turn into debian apt
2022-12-03 18:17:53 <psykose> should've foreseen this
2022-12-03 18:18:09 <Newbyte> yes, this one-line code change is a hostile takeover to make apk interactive by default
2022-12-03 18:18:12 <Newbyte> you figured out my evil plan
2022-12-03 18:18:13 <Newbyte> noooo
2022-12-03 18:18:34 <panekj> psykose: it's literally apt equivalent
2022-12-03 18:18:42 <psykose> are you high on crack
2022-12-03 18:18:50 <panekj> no, but you seem to be
2022-12-03 19:12:17 <q66> i may have made a terrible mistake
2022-12-03 19:12:45 <panekj> ?
2022-12-03 19:13:34 <q66> a simple suggestion and it turns into 20 minutes of pointless arguing 
2022-12-03 19:14:39 <psykose> i fell for the bait as always
2022-12-03 19:14:48 <panekj> I would argue regardless if you suggested something or not, so don't beat yourself for that
2022-12-03 19:16:45 <q66> Newbyte: would the changes really be that substantial
2022-12-03 19:16:58 <q66> interactive effectively only affects stuff that changes world
2022-12-03 19:17:37 <q66> or well, changes the filesystem, since iirc it also affects fix which does not change world
2022-12-03 19:17:44 <q66> still, only add/del/fix
2022-12-03 19:18:01 <psykose> afaik it's mostly for pmbootstrap and there is an entire apk wrapper in there
2022-12-03 19:18:31 <psykose> so it's 2 lines of code to throw in the yes | override
2022-12-03 19:18:50 <q66> also imo being explicit about the command being potentially interactive is better than setting some env var somewhere which may not be obvious at all 
2022-12-03 19:18:53 <psykose> could be wrong though, didn't follow any of the discussions
2022-12-03 19:19:10 <psykose> just saw them enable it and someone saying everything broke a day later on the issue tracker
2022-12-03 19:20:34 <psykose> yeah. would say --no-interactive would be better as the override so it's obvious where disabled
2022-12-03 19:20:35 <psykose> shrug
2022-12-03 19:21:31 <liske> 
ACTION is just wondering if apk behave non-interactive if stdin is closed before exec

2022-12-03 19:21:58 <q66> yeah I was thinking about that too 
2022-12-03 19:22:25 <psykose> something something easy to test
2022-12-03 19:22:56 <psykose> it just closes xD
2022-12-03 19:23:06 <psykose> assumes n it seems
2022-12-03 19:23:09 <liske> so it is non-interactive
2022-12-03 19:23:12 <liske> \o/
2022-12-03 19:23:18 <psykose> sure, with an automatic fail
2022-12-03 19:23:28 <liske> ;)
2022-12-03 19:23:31 <psykose> :)
2022-12-03 19:35:54 <Newbyte> psykose: I was thinking about the risk of this being confusing. My idea was to print a warning that the environment variable it set, but changing it to something like --non-interactive could work too. as for pmbootstrap, I looked myself and there are more than 2 lines that would have to be changed I think sadly.
2022-12-03 19:45:26 <Newbyte> (but sure, it would definitely be feasible)
2022-12-03 22:57:15 <skarnet> please default to non-interactive if stdin isn't a terminal
2022-12-04 01:19:12 <elibrokeit> Newbyte: when testing apk via Meson, I don't quite understand why you have an issue with changing invocation of apk? Setting an LD_PRELOAD variable is hardly going to help you if you modify the sources for the apk binary instead of the executable.
2022-12-04 01:19:53 <elibrokeit> Fortunately none of this matters, simply add $PWD/src to your $PATH and have `apk` find the one in the build directory first
2022-12-04 01:20:56 <elibrokeit> ... or you could also do `meson devenv my-prog-that-uses-apk` which will be run inside an environment that is set up to fully enable the uninstalled version of the built project (sets PATH, but also sets lots of other things)
2022-12-04 09:39:29 <Newbyte> elibrokeit: I had misunderstood what psykose suggested me to do
2022-12-04 09:40:17 <Newbyte> I thought she meant that I just could preload the custom-built libapk and that would somehow work with the installed apk
2022-12-04 16:05:25 <vkrishn> psykose: newsboat date issue goes back to v3.13
2022-12-04 16:05:35 <psykose> i'm aware
2022-12-04 21:26:37 <vkrishn> shouldn't alpine-minirootfs...tar.gz also include .alpine-release file ?
2022-12-04 21:29:48 <minimal> vkrishn: you mean /etc/alpine-release ?
2022-12-04 21:30:27 <vkrishn> yeah, found it, was searching .alpine-release, thanks
2022-12-05 13:43:50 <chereskata> Hi,
2022-12-05 13:45:57 <chereskata> i have had a nice small conversation with Bart on e-mail, that started from a false outdated flag by me on pkgs.alpinelinux and ended in a suggestion of mine to use repology in addition to Anitya to flag outdated packages 
2022-12-05 13:46:15 <chereskata> some packages were outdated and i had to flag them manually while looking them up in repology
2022-12-05 13:47:02 <chereskata> is it worth it to write some small piece of webcrawler to autoflag packages with repology info?
2022-12-05 13:47:50 <psykose> are you the one that flags things with Dear x, ..
2022-12-05 13:47:56 <chereskata> yes
2022-12-05 13:48:28 <psykose> half of them aren't even correct or are like 2 hours after the project even updates
2022-12-05 13:49:09 <psykose> gnupg 2.3.8? xD
2022-12-05 13:49:17 <chereskata> i'll look better next time
2022-12-05 13:49:53 <chereskata> https://repology.org/project/gnupg/versions
2022-12-05 13:50:28 <psykose> repology is probably the worst packaging source that exists out there
2022-12-05 13:50:43 <chereskata> good to know, sorry for the noise
2022-12-05 13:50:59 <psykose> in general though, the automatic flagging already on the list is via https://release-monitoring.org/
2022-12-05 13:51:10 <psykose> if you add packages to that they would get flagged
2022-12-05 13:51:58 <psykose> so for instance there is https://release-monitoring.org/project/1764/ , and there is a mapping to the alpine pkgname
2022-12-05 13:52:14 <psykose> not always correct (and our integration for ignoring pre-releases is still broken), but it's alright
2022-12-05 13:52:42 <chereskata> that is really cool!
2022-12-05 13:53:32 <chereskata> Thank you!
2022-12-05 13:53:56 <psykose> for plenty of projects they are already there, just needs a mapping
2022-12-05 13:54:08 <psykose> to modify the lists you need a fedora account, should be easy to find via the links
2022-12-05 13:54:24 <chereskata> after taking a brief look over the docs, it does not look too complicated
2022-12-05 13:54:37 <psykose> usual email signup, nothing fancy
2022-12-05 13:55:14 <chereskata> Thanks again
2022-12-05 20:16:25 <ikke> psykose: lol, sparkling plaintext
2022-12-05 20:16:29 <psykose> :)
2022-12-05 20:17:20 <ikke> why is nextcloud still using it..
2022-12-05 20:17:41 <psykose> didn't read too much but i'd guess it was for on-disk encryption only? shrug
2022-12-05 20:17:53 <psykose> i'd guess it's not even on by default and quite redundant
2022-12-05 20:18:00 <psykose> but i really didn't check any of that or ever used nextcloud
2022-12-05 20:18:04 <psykose> as for why, probably legacy
2022-12-05 20:18:09 <psykose> torrent encryption is rc4 too
2022-12-05 20:18:24 <psykose> that one is harder (impossible) to change though, as it's the standard
2022-12-05 20:22:34 <psykose> or was it? my memory sucks
2022-12-05 20:31:47 <minimal> sparkling plaintext indeed. i haven't used Nextcloud so far but def not impressed they're still relying on RC4
2022-12-06 13:32:20 <markand> anyone one !40071
2022-12-06 13:32:23 <markand> -e
2022-12-06 15:54:36 <bl4ckb0ne> psykose: that plugdev stuff is annoying
2022-12-06 15:54:59 <bl4ckb0ne> but i seem to be the only one bothered by it, i can just live with the rule added by myself in udev
2022-12-06 15:59:56 <psykose> i mean i'd prefer if you found out the real issue :p
2022-12-06 16:00:08 <psykose> sounds difficult though
2022-12-06 16:17:42 <bl4ckb0ne> yeah
2022-12-06 16:17:49 <bl4ckb0ne> im lacking time and motivation
2022-12-06 16:18:02 <bl4ckb0ne> imo pmos could just yeet the uaccess tag on their shit
2022-12-06 18:36:31 <ngortheone> Hi there, is it possible to find html/css + backend sources of https://pkgs.alpinelinux.org/packages ? I want to build a similarly looking website, thanks!
2022-12-06 18:37:07 <ikke> https://gitlab.alpinelinux.org/alpine/infra/aports-turbo
2022-12-06 18:41:45 <ngortheone> wow, it is also written in Lua, how cool is that?!
2022-12-06 18:42:05 <ngortheone> thanks ikke!
2022-12-06 23:31:06 <omni> pretty cool
2022-12-07 14:15:41 <omni> so, looking at !42109 it seems like something changed in alpine for x86 that had that test fail?
2022-12-07 21:56:23 <bl4ckb0ne> is it normal that the latest 2 pipelines on master were canceled?
2022-12-07 21:56:40 <bl4ckb0ne> the packages have not been built
2022-12-07 21:56:46 <psykose> which ones
2022-12-07 21:57:29 <psykose> if you mean something like https://gitlab.alpinelinux.org/alpine/aports/-/pipelines/146100 those don't build any packages
2022-12-07 21:57:40 <psykose> builders are separate
2022-12-07 21:57:44 <psykose> and rebuilding a bunch of go stuff
2022-12-07 21:57:58 <psykose> that would be https://build.alpinelinux.org/
2022-12-07 21:58:07 <bl4ckb0ne> ah i was looking for the url
2022-12-07 21:58:09 <bl4ckb0ne> ty
2022-12-07 21:58:29 <bl4ckb0ne> ill wait overnight then
2022-12-07 21:58:31 <ikke> if you care enough, you can also hang around in #alpine-commits
2022-12-07 21:58:37 <bl4ckb0ne> ci still red :<
2022-12-07 21:58:43 <bl4ckb0ne> my*
2022-12-07 21:58:56 <psykose> for x86_64 i'd give it another 2 hours
2022-12-07 21:59:09 <psykose> or three
2022-12-07 23:05:44 <bl4ckb0ne> ill make it 12 to be sure its there
2022-12-08 17:27:54 <omni> seems like Xen 4.17 will be released on monday
2022-12-09 01:05:09 <omni> so.. how do we make the linter not care about backticks in comments?
2022-12-09 01:11:44 <psykose> there's like 15 similarish issues open for the same thing in it
2022-12-09 01:11:45 <psykose> xd
2022-12-09 01:16:13 <omni> where? I could open a 16th
2022-12-09 01:16:42 <psykose> shellcheck https://github.com/koalaman/shellcheck
2022-12-09 01:18:07 <omni> aah
2022-12-09 01:20:39 <omni> so.. what warran a release for that project? someone or something to dedicate it to?
2022-12-09 01:59:37 <panekj> just get rid of shellcheck completely
2022-12-09 02:06:28 <minimal> omni: if the backticks is triggering a specific shellcheck warning/error then the linter could pass shellcheck the "-e ~SC????" option to tell it to ignore that check
2022-12-09 03:11:01 <elibrokeit> shellcheck is an excellent tool for giving hints to new users, but if you actually know what you're doing then it rarely if ever tells you something you care about. Many of its warnings don't actually tell you to make a change -- only that you should evaluate the code and verify if XXX problem is a concrete issue there.
2022-12-09 05:46:38 <ayakael> https://gitlab.alpinelinux.org/alpine/aports/-/commit/b07f88d2e613262e0673f84adb88d2b50b0774cf Anyway to neutralize --as-needed now being default? I tried setting -Wl,--no-as-needed to no avail. It breaks build of dotnet7 on s390x.
2022-12-09 05:48:11 <psykose> if as-needed breaks the build then the build is broken
2022-12-09 05:50:22 <ayakael> Well, -lucontext worked before and now C code that should find ucontext symbols isn't finding it.
2022-12-09 05:52:45 <psykose> broken things worked before yes
2022-12-09 05:55:17 <psykose> if you need to link libucontext to a vendored libunwind then you must link it to the vendored libunwind
2022-12-09 05:55:38 <psykose> target_link_libraries(coreclrpal ucontext) does not look like libunwind
2022-12-09 05:55:44 <psykose> it looks like something else
2022-12-09 05:55:50 <psykose> so it gets discarded and then it's missing
2022-12-09 05:56:14 <psykose> you have to pass it to the actual libunwind link step
2022-12-09 05:56:47 <psykose> this would have already been broken if this very good build system actually read the ldflags
2022-12-09 05:59:34 <ayakael> I appreciate the pointers on best practice. It is also my understanding that this build system has many problems, so there's really no need to rub it in.
2022-12-09 06:09:14 <psykose> apologies, not really how i intended that to sound like
2022-12-09 06:15:55 <ikke> omni: "The search for the matching backquote shall be satisfied by the first unquoted non-escaped backquote; during this search, if a non-escaped backquote is encountered within a shell comment, a here-document, an embedded command substitution of the $(command) form, or a quoted string, undefined results occur"
2022-12-09 06:15:59 <ikke> https://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_06_03
2022-12-09 06:16:16 <ikke> So apparently it's on purpose that it warns about it
2022-12-09 06:20:58 <psykose> hmm
2022-12-09 06:21:03 <psykose> so not even comments are safe eh
2022-12-09 06:22:42 <ikke> It doesn't seem the common shells seems to do anything with it
2022-12-09 06:22:58 <psykose> well, it would be quite the mess :/
2022-12-09 06:23:01 <ikke> tested # `touch foo`, and on none of the common shells the file foo appeared
2022-12-09 06:23:24 <psykose> i think it's more like
2022-12-09 06:23:28 <psykose> # `thing
2022-12-09 06:23:35 <psykose> and then that paired with some other stuff can do funny things
2022-12-09 06:23:39 <ikke> right
2022-12-09 06:23:53 <psykose> just a # `abc` should always be safe
2022-12-09 06:24:06 <psykose> i just find it strange that # <anything> can be dangerous
2022-12-09 06:24:26 <ikke> yes
2022-12-09 06:24:32 <ikke> unexpected
2022-12-09 07:31:30 <skarnet> "undefined results occur" only means "historically, there are very stupid shells that do very stupid things"
2022-12-09 07:32:14 <skarnet> going forward I very much hope that comments would take precedence above anything else
2022-12-09 08:23:13 <ikke> Nod
2022-12-09 08:56:48 <liske> I have a install="$pkgname.pre-install" in APKBUILD and abuild complains "You should not have $install in source"
2022-12-09 08:58:16 <liske> but the Alpine wiki doesn't seem to depreciate install=
2022-12-09 08:59:05 <liske> Should the warning be ignored?
2022-12-09 09:08:22 <ikke> liske: did you include those files in source= as well? 
2022-12-09 09:08:39 <psykose> you don't put it in source at all
2022-12-09 09:08:41 <psykose> just install=
2022-12-09 09:09:00 <liske> they are in $source
2022-12-09 09:09:05 <psykose> then remove them from source
2022-12-09 09:11:21 <liske> ooops, I did misunderstood the warning
2022-12-09 09:11:48 <liske> I was reading $install should not be in APKBUILD *facepalm*
2022-12-09 09:11:53 <liske> sry
2022-12-09 09:15:06 <ikke> Np
2022-12-09 11:14:37 <omni> ikke: so not a bug, then, I'll just use single quotes instead
2022-12-09 12:37:44 <eddsalkield> would someone be around to review/merge my patch?: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41202
2022-12-09 15:15:51 <chereskata> Hello,
2022-12-09 15:17:57 <chereskata> i have noticed, that /etc/pam.d/base-password contains nullok as well as md5 
2022-12-09 15:19:03 <chereskata> both could pose a security risk, especially when the root filesystem is unencrypted or /etc/shadow world readable (not the case) 
2022-12-09 15:19:20 <chereskata> as i have seen, debian use yescrypt as of Debian 11
2022-12-09 15:20:01 <chereskata> https://www.debian.org/releases/bullseye/amd64/release-notes/ch-information.en.html#pam-default-password
2022-12-09 15:39:02 <minimal> chereskata: if the root filesystem is uncrypted then it could be accessed by booting Linux off USB and mounting it, or removing the drive and using it on another computer etc...
2022-12-09 15:39:41 <chereskata> I am very aware of the usefullness of encryption
2022-12-09 15:40:01 <minimal> I'd agree that md5 should be replaced with something stronger
2022-12-09 15:41:55 <chereskata> i tried yescrypt, but it gave me a memory error on passwd
2022-12-09 15:42:03 <chereskata> so i settled with sha512 for now
2022-12-09 15:42:28 <minimal> as for "nullok", that's really down to what typical "out-of-the-box" use case is catered for - some people may want passwordless accounts for local access (distinct from restricting SSH to keys/certs only) - should Alpine cater for that by default? or severely locked down by default? (actually PAM is not installed by default in Alpine)
2022-12-09 15:43:09 <minimal> chereskata: remember Alpine uses musl - some strong(er) password encryption methods may not work with musl
2022-12-09 15:44:16 <chereskata> minimal: as i think alpine caters to advanced users and those who want to become such (like me), i think it is easier to loosen the security belt, than to tighten it up (time issues)
2022-12-09 15:44:57 <chereskata> Thanks for the clarification
2022-12-09 15:45:19 <minimal> chereskata: well at the very least there's a distinction as to who Alpine caters for and who uses Alpine
2022-12-09 15:45:36 <chereskata> in what means?
2022-12-09 15:45:53 <minimal> i.e. lots of people use Alpine as a container base image yet I don't believe Alpine has ever specifically catered for that, it just happened
2022-12-09 15:46:27 <chereskata> that is a good point
2022-12-09 15:47:35 <minimal> so the "needs" of Alpine a server OS, as a container base, and as a desktop are in some ways distinct an even possibly conflicting
2022-12-09 15:48:15 <chereskata> yes, that is true. Reminds me a lot of debian work machines, that run dbus for basic UI less tasks
2022-12-09 15:52:05 <minimal> I agree that the linux-pam config probably should be looked over and tweaked/tightened up, I just don't kno how many (5%? less?) Alpine machines actually make use of PAM
2022-12-09 15:52:28 <chereskata> The desktop ones, at least these with elogind
2022-12-09 15:52:47 <chereskata> probably downstream pmOS has more clients with it
2022-12-09 15:53:21 <chereskata> Personally i run gnome
2022-12-09 15:54:07 <minimal> I haven't run Alpine as a desktop yet
2022-12-09 15:55:32 <elly> I am using it on my laptop currently
2022-12-09 15:55:39 <elly> but without pam
2022-12-09 15:56:44 <DavidHeidelberg[m]> minimal: but I think your nickname makes running alpine on desktop a requirement
2022-12-09 15:56:53 <chereskata> +1
2022-12-09 15:57:19 <chereskata> elly: i suppose you are using a more "slim" wm?
2022-12-09 15:57:44 <minimal> wouldn't running it via just a text console be more realistically minimal? ;-)
2022-12-09 15:59:08 <minimal> which is when the terms "server" and "desktop" get blured, it its a PC on my desk but with only a text console is it a desktop? ;-)
2022-12-09 16:04:42 <elly> I am using i3
2022-12-09 16:04:48 <elly> it is working well for me so far
2022-12-09 16:06:59 <chereskata> how is your stance on the insecurities regarding unrestricted  access of screen data by applications?
2022-12-09 16:11:35 <elly> like for my personal use case? I don't worry about it
2022-12-09 16:19:21 <ikke> eddsalkield: will check it in a bit
2022-12-09 17:10:41 <eddsalkield> ikke: thanks
2022-12-10 15:58:10 <psykose> https://img.ayaya.dev/Y8FvjfF00pSB
2022-12-10 15:58:12 <psykose> so close yet so far
2022-12-10 16:38:08 <psykose> ah, right, no radeonsi support yet
2022-12-10 18:24:41 <bl4ckb0ne> no rusticle?
2022-12-10 18:34:13 <psykose> works
2022-12-10 18:34:17 <psykose> well, kinda
2022-12-10 18:34:26 <psykose> needed a build fix and it aborts at runtime
2022-12-10 18:45:24 <bl4ckb0ne> you reported the issue?
2022-12-10 18:48:38 <psykose> it's already fixed 2 weeks ago
2022-12-10 18:48:40 <psykose> just not in 22.3.0
2022-12-10 18:48:59 <psykose> didn't see until i also fixed it with some help in dri-devel of course
2022-12-10 18:49:00 <psykose> as always
2022-12-10 18:49:04 <psykose> now more things to fix :)
2022-12-10 22:30:49 <anjan_> Hi, I need help with apk. So pipewire-alsa "provides" pulseaudio-alsa.
2022-12-10 22:31:00 <anjan_> I want a package to specifically install pulseaudio-alsa
2022-12-10 22:31:06 <anjan_> pipewire-alsa will not do.
2022-12-10 22:31:20 <anjan_> Can I do that in the package build Im writing
2022-12-10 22:31:38 <panekj> just require pulseaudio-alsa?
2022-12-10 22:32:19 <anjan_> you would think that would work
2022-12-10 22:32:21 <psykose> what package is it
2022-12-10 22:32:32 <anjan_> psykose: it's a package in postmarketos
2022-12-10 22:32:43 <psykose> nmeum: urg, i know you love sigign, so make 4.4 now gets the same sigpipe issue fe27b8e5bc546480ad4fc58df0958d8f9484ba25
2022-12-10 22:32:48 <psykose> it seems they are also doing fun stuff now
2022-12-10 22:33:10 <panekj> I don't think
2022-12-10 22:33:11 <anjan_> diff: https://paste.sr.ht/~anjan/1bbfeadd5c01a79f9a8170652f15166c91e9396c and package https://gitlab.com/postmarketOS/pmaports/-/blob/master/main/sxmo-common/APKBUILD
2022-12-10 22:33:37 <anjan_> psykose: I want to be able to have a package depend on sxmo-common-pulse and it installs pulseaudio stuff and removes pipewire.
2022-12-10 22:33:55 <psykose> nmeum: that's the first failure, probably won't be very common or anything, but i wonder if this is something that needs more looking at (already reported in https://savannah.gnu.org/bugs/index.php?63307 , but)
2022-12-10 22:34:02 <panekj> btw. it's abuild not apk
2022-12-10 22:35:25 <psykose> the subpackage already does depends=pulseaudio-alsa !pipewire-alsa so it has the dependency constraints you want
2022-12-10 22:36:20 <anjan_> psykose: ok, I have that package installed (confirmed with doas apk -Ua upgrade). doas apk policy pulseaudio-alsa shows that it's not installed.
2022-12-10 22:36:36 <anjan_> `doas apk policy pipewire-alsa` shows pipewire-alsa is installed
2022-12-10 22:36:55 <psykose> apk info -R sxmo-common
2022-12-10 22:37:00 <psykose> and post it
2022-12-10 22:37:22 <psykose> sxmo-common-pulse too
2022-12-10 22:38:20 <anjan_> gimme a sec
2022-12-10 22:38:26 <anjan_> my pastebin script broke =(
2022-12-10 22:38:30 <psykose> :)
2022-12-10 22:41:45 <anjan_> psykose: https://paste.sr.ht/~anjan/9fa6985e40702762b36c05cb17b1979c7324539e
2022-12-10 22:42:31 <anjan_> psykose: https://paste.sr.ht/~anjan/e0c59e1bafa9115553586f37a9665a5d29329200
2022-12-10 22:42:48 <psykose> apk policy sxmo-common-pulse
2022-12-10 22:43:32 <anjan_> https://paste.sr.ht/~anjan/088c371cb71030c8bc3fa04937ac9f8cce515964
2022-12-10 22:43:42 <psykose> you have -r1 installed, and it's the first in the list
2022-12-10 22:43:46 <psykose> the one without the constraints
2022-12-10 22:44:22 <psykose> now you can do `apk add sxmo-common-pulse=1.11.0-r2` to get the error as to why it doesn't upgrade
2022-12-10 22:45:40 <anjan_> psykose: https://paste.sr.ht/~anjan/a536c074aecf0a86ad0586b60cb0bc7da5e37223
2022-12-10 22:46:12 <psykose> you have pipewire-alsa in your world and you're trying to add a package that does !pipewire-alsa
2022-12-10 22:46:32 <anjan_> Thank you. Ill look into it =P
2022-12-10 22:46:51 <anjan_> I appreciate explaining errors to me. Ive been using alpine for 2 years and Im still a beginner =P
2022-12-10 22:47:15 <anjan_> ok
2022-12-10 22:47:17 <anjan_> it's working
2022-12-10 22:48:11 <anjan_> thank you for all your help as always psykose. I learned alot. Ill take notes.
2022-12-10 22:48:11 <psykose> it's allowed to do `apk add !x` with x already installed and it removes it, but not in dependencies
2022-12-10 22:48:42 <anjan_> oh that makes much more sense
2022-12-10 22:49:09 <psykose> (because the commandline modifies the world file, so that can remove something, but nothing else can modify it so it would conflict)
2022-12-10 22:50:54 <psykose> in general this is another reason i don't like virt packages very much- with a bit of experience it's not hard to debug these failures (though that absolutely does not fly for random users, which is terrible), but there is no error reporting at all when something cannot upgrade
2022-12-10 22:51:30 <psykose> so minimising the chance of even getting into a state where some upgrade is skipped because of wrong dependencies all over the place (especially when its vulnerable to something already in someones world file) is a bit easier to make it consistent
2022-12-10 22:51:30 <psykose> shrug
2022-12-10 22:53:46 <psykose> realistically it would be nice if there was error reporting for a skipped upgrade that is not overruled via a version constraint (i.e. no error if x<5 and 6 is available), since in general there aren't cases where it's correct in the practical sense to not upgrade
2022-12-10 22:53:50 <psykose> the exception is of course downgrades
2022-12-10 22:53:57 <psykose> and then the implementation becomes quite difficult
2022-12-10 22:54:34 <psykose> in any case i can't implement any of that or just fix it all so i go for the simpler packaging solution :)
2022-12-10 23:05:53 <craftyguy> psykose: a talk, blog, wiki edit or something on these tricks to avoid making virtual packages (and other useful knowledge of abuild/apk you have) would be amazing :D
2022-12-10 23:06:06 <psykose> me, a talk?!
2022-12-10 23:06:22 <craftyguy> because I also find it tempting to create a virt package when there's probably a much better way
2022-12-10 23:06:29 <psykose> hah
2022-12-10 23:06:44 <psykose> well to be clear to the above: i'm not really pointing to a better way
2022-12-10 23:07:18 <panekj> psykose: a podcast
2022-12-10 23:07:20 <craftyguy> ah, well an alternative in this case. even if the tradeoffs are different, it might be "better" in some ways
2022-12-10 23:07:43 <psykose> in general it's more about the-list-of-packages-you-want-is-documented-somewhere, or something, but purely a virtual that doesn't do anything except pull in a set from something works alright
2022-12-10 23:07:52 <psykose> i.e. the kde ones, the gnome one, 
2022-12-10 23:09:16 <psykose> i don't really think there's a hard and fast rule though, hm
2022-12-10 23:09:31 <psykose> i guess it's mostly having to know all of the dependencies in the distro and how they're laid out
2022-12-10 23:09:35 <psykose> that especially goes for install_if
2022-12-10 23:16:46 <psykose> in this specific case, it probably makes more sense to have a pipewire/pulseaudio generic thing for the entire distro (and not here at all), and then if you need a specific override (like an extra configuration file for pulseaudio), you can install_if="$pkgname=$pkgver-r$pkgrel distro-pulseaudio", and the bluetooth thing can !pulse the other way around
2022-12-10 23:16:54 <psykose> but there's probably 10 edgecases i'm not thinking of :p
2022-12-10 23:17:55 <psykose> (and mostly that's just to not have these 20 redundant packages for 'system uses pulseaudio' and 'system uses pipewire' all over the place, when it's the exact same dependencies)
2022-12-10 23:18:36 <psykose> but in the end, alpine doesn't have this, and people mostly figure out how to install the 3-4 things anyway. unless i just don't hear about all the others
2022-12-10 23:26:23 <psykose> (but yeah, there's also nothing i can really suggest as i haven't spent the 50 hours thinking over whats the best way to do all of xyz with some requirements with some pmos packaging. but you can always ping me for help with specific cases like above if something doesn't work :) )
2022-12-11 13:56:12 <psykose> nmeum: ah, also fyi binutils-gold isn't needed for go anymore on arm, as the actual bfd issue it was for has been fixed for over a year https://sourceware.org/bugzilla/show_bug.cgi?id=19962
2022-12-11 13:56:27 <psykose> it would need a patch to remove the check, but gold isn't required otherwise
2022-12-11 14:18:57 <nmeum> can look into later today but if we need a patch to get rid of it then I would rather keep it for now
2022-12-11 14:19:53 <psykose> mostly it's an honor to rm -rf google code for fun
2022-12-11 14:19:55 <psykose> but also yes
2022-12-11 15:35:23 <omni> https://gitlab.alpinelinux.org/-/snippets/561
2022-12-11 15:44:37 <psykose> ah this libcxx issue again
2022-12-11 15:44:44 <psykose> i am getting quite tired of this apk edgecase :)
2022-12-11 15:47:32 <psykose> omni: c58214f0c7837c341d0976a7225fbf4e8a53585b
2022-12-11 18:01:37 <sam_> psykose: is this a file->dir thing
2022-12-11 18:01:39 <sam_> for __string
2022-12-11 18:01:44 <psykose> yes
2022-12-11 18:01:55 <psykose> apk does not handle that in 1 transaction you just get a replace failure
2022-12-11 18:01:59 <sam_> same for us
2022-12-11 18:02:02 <psykose> : )
2022-12-11 18:02:04 <sam_> we tried https://reviews.llvm.org/D139270 for a similar thing and nobody really got what we meant
2022-12-11 18:02:28 <psykose> tbf it just seems like a niche implementation detail to me
2022-12-11 18:02:35 <psykose> i really can't blame random projects for it
2022-12-11 18:02:48 <psykose> python gets the same issue
2022-12-11 18:03:07 <psykose> anything built with old setuptools or no-setuptools using something 60+ goes from .egg-info to .egginfo/
2022-12-11 18:03:12 <psykose> .egg-info/*
2022-12-11 18:03:19 <psykose> and without an upgrade or something, same issue
2022-12-11 18:04:39 <psykose> (at least that one has a hack lol, as anything using 'setuptools' is quite easy to just hack into pep517 without a pyproject at all and then you get a dist-info instead)
2022-12-11 18:04:42 <psykose> boom, rename solved
2022-12-11 18:57:41 <sam_> i don't blame them, it's just also an easy workaround for a project to implement given a few downstreamsh it it
2022-12-12 06:38:28 <ikke> linux 6.1 is out
2022-12-12 10:21:22 <omni> I'm having issues with xen 4.17.0 and cpupools where I cannot start DomUs, trying to figure it out
2022-12-12 10:21:50 <omni> heads-up to anyone running xen on edge
2022-12-12 11:13:58 <psykose> hurra for mglru
2022-12-12 11:14:03 <psykose> another nice linux release indeed
2022-12-12 11:19:10 <ikke> https://www.phoronix.com/news/MGLRU-In-Linux-6.1
2022-12-12 11:19:45 <ikke> so improvements on memory management?
2022-12-12 11:22:09 <psykose> yeah, just yet another low level rework in some subsystem
2022-12-12 11:22:23 <psykose> especially nice on all those <4gb vms
2022-12-12 11:57:14 <tomzy_0> hello to all
2022-12-12 11:57:53 <tomzy_0> are there any tools in apline linux which would allow me to prepare image that supports dual rootfs partition layout?
2022-12-12 11:58:45 <tomzy_0> the goal here would be to later use some update system with A/B dual image mechanism
2022-12-12 12:00:27 <tomzy_0> e.g. use this https://rauc.readthedocs.io/en/latest/scenarios.html#symmetric-root-fs-slots
2022-12-12 12:05:47 <panekj> if you could not spam in all channels
2022-12-12 12:05:51 <panekj> that would be great
2022-12-12 15:42:30 <chereskata> Hi, why does this dependency exist? Shouldn't pipewire-pulse replace it?
2022-12-12 15:42:45 <chereskata> apk del --purge pulseaudio
2022-12-12 15:42:53 <chereskata> World updated, but the following packages are not removed due to:
2022-12-12 15:43:02 <chereskata>   pulseaudio: pipewire-pulse gnome-apps-core gnome
2022-12-12 15:43:19 <chereskata> As i am not sure, i dont want to spam the bugtracker
2022-12-12 15:43:50 <psykose> pipewire-pulse provides= pulseaudio but you don't actually have it
2022-12-12 15:44:05 <chereskata> Ah okey nice :). Thanks for the clarification
2022-12-12 16:16:24 <chereskata> Could someone merge my aports change? https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/42262
2022-12-12 16:16:28 <chereskata> Thank you :)
2022-12-12 16:18:31 <psykose> your fork is not public
2022-12-12 16:36:28 <chereskata> I fixed it
2022-12-12 16:59:57 <omni> I got help from j_gross of #xen
2022-12-12 17:00:11 <omni> !42278 is the hotfix
2022-12-12 17:00:51 <omni> but I'd like to look more at what gentoo have done with their xen openrc scripts the past few years
2022-12-12 17:01:30 <omni> https://gitweb.gentoo.org/repo/gentoo.git/tree/app-emulation/xen-tools/files
2022-12-12 17:10:04 <omni> perhaps better to just get that in asap
2022-12-12 17:17:39 <PabloCorreaGomez[m]> ikke Cogitri: any chance you can have another look at this one? https://gitlab.alpinelinux.org/alpine/infra/aports-qa-bot/-/merge_requests/38 Should be ready from my point of view
2022-12-12 19:01:58 <omni> please merge !42278, I'll take a deeper look at things in !42283
2022-12-12 19:04:09 <psykose> are you sure it's correct? i.e. what it replaces is definitely obsoleted and the new thing does the same but new and working
2022-12-12 19:14:05 <omni> I'm running with it now, before I couldn't create DomUs because the Cpupool was (null)
2022-12-12 19:14:42 <psykose> yeah, i'm just wondering what it really does
2022-12-12 19:14:54 <psykose> i know one works and the other doesn't :)
2022-12-12 19:17:00 <omni> http://xenbits.xen.org/gitweb/?p=xen.git;a=blob;f=tools/helpers/xen-init-dom0.c;hb=refs/tags/RELEASE-4.17.0
2022-12-12 19:18:01 <omni> <j_gross> omni: there was a change in 4.17 requiring that xen-init-dom0 is called during system start in order to have the cpupool info initialized. In Xen 4.16 the mapping of poolid 0 -> "Pool-0" was hard wired into libxl.
2022-12-12 19:19:31 <omni> gentoo does this since five years https://gitweb.gentoo.org/repo/gentoo.git/tree/app-emulation/xen-tools/files/xenstored.initd-r1
2022-12-12 19:20:50 <psykose> makes sense
2022-12-12 19:21:06 <psykose> thanks
2022-12-12 21:25:53 <omni> no, thank you
2022-12-13 12:30:51 <handlerug> is there a reason why hare{,-ssh,-json} packages were disabled on aarch64? (https://gitlab.alpinelinux.org/alpine/aports/-/commit/6416744f261a1f5759f6c4a7e3715949e8a53114) cc psykose
2022-12-13 12:31:00 <handlerug> all of them build just fine on my system
2022-12-13 12:36:17 <ncopa> handlerug: it probably did not build at some point
2022-12-13 12:39:36 <ikke> Around the release hare was giving issues 
2022-12-13 16:31:57 <ncopa> $ uname -a
2022-12-13 16:31:58 <ncopa> Linux ncopa-desktop 6.1.0-0-lts #1-Alpine SMP PREEMPT_DYNAMIC Tue, 13 Dec 2022 15:49:51 +0000 x86_64 GNU/Linux
2022-12-13 16:33:29 <psykose> not official yet :p
2022-12-13 16:33:42 <psykose> did you enable LRU_GEN
2022-12-13 16:34:07 <ncopa> nope, i just used the current config + defconfig
2022-12-13 16:34:45 <ncopa> # CONFIG_LRU_GEN is not set
2022-12-13 16:35:21 <psykose> unfortunate
2022-12-13 16:35:24 <ncopa> i suppose i should enable that
2022-12-13 16:36:27 <psykose> there was some more stuff 5.15->6.1 range but i forgot most of them
2022-12-13 16:36:31 <psykose> i guess the usual slew of drivers
2022-12-13 16:36:37 <psykose> bunch of intel/amd platform stuff
2022-12-13 16:36:58 <psykose> amd_pstate etc
2022-12-13 16:37:17 <ncopa> we pick up whatever is defconfig
2022-12-13 16:37:46 <Ermine> psykose: is this MGLRU?
2022-12-13 16:38:04 <Ermine> (LRU_GEN)
2022-12-13 16:38:24 <psykose> the majority of things are never in defconfig
2022-12-13 16:38:26 <psykose> Ermine: yes
2022-12-13 16:38:51 <psykose> it's actually a little funny what is or isn't, given entire intel chipsets for laptops default to =n half the time
2022-12-13 16:39:01 <psykose> but complete random ethernet driver for one pcie card defaults to =m or whatever
2022-12-13 16:39:06 <psykose> not very useful
2022-12-13 16:50:25 <q66> so what's the benefit of CONFIG_LRU_GEN
2022-12-13 16:50:44 <psykose> new feature go fast big feature yes
2022-12-13 16:50:51 <psykose> but it's mostly in low memory situations
2022-12-13 16:52:18 <q66> <ncopa> nope, i just used the current config + defconfig
2022-12-13 16:52:27 <q66> does that mean you have to actually manually figure out what new configs appeared
2022-12-13 16:53:02 <q66> what i do for kernel builds is just keep the actual dotconfig, and then give it to a new kernel, oldconfig and have it generate a new one
2022-12-13 16:53:08 <q66> then it asks me for everything that has changed
2022-12-13 17:06:37 <ncopa> I used to do that before
2022-12-13 17:09:20 <q66> imo it's the only approach that really works
2022-12-13 17:09:31 <q66> defconfigs don't always contain what you want (usually they are missing drivers at very least)
2022-12-13 17:09:43 <q66> and it gets hard to keep track of what really changed in the end
2022-12-13 17:10:49 <ncopa> I do keep track of what differs from defconfig
2022-12-13 19:44:35 <chereskata> As i have posted here: https://gitlab.alpinelinux.org/alpine/aports/-/issues/14421 , i can not share my screen on wayland. where should i start looking for to fix pipewire screensharing with wireplumber?
2022-12-13 19:46:59 <psykose> it's most likely just a pipewire issue
2022-12-13 19:47:08 <psykose> they fixed some regression in .62 but it's still broken
2022-12-13 19:53:55 <omni> have you tried 'screen -x'?
2022-12-13 19:53:57 <omni> 
ACTION runs

2022-12-13 19:56:06 <chereskata> omni: is this geared towards my question?
2022-12-13 19:56:25 <chereskata> psykose: should i report it upstream?
2022-12-13 19:56:53 <psykose> if you downgrade to .59/.60 and it begins working, sure
2022-12-13 20:00:56 <chereskata> is it possible to download precompiled apks?
2022-12-13 20:04:58 <psykose> as opposed to non-compiled apks?
2022-12-13 20:08:23 <panekj> just use this beauty https://github.com/wnayes/electron-wm /s
2022-12-13 20:17:12 <bl4ckb0ne> 
ACTION gags

2022-12-13 20:18:06 <chereskata> psykose: What would work when i revert the aports repo to before pipewire had been updated  
2022-12-13 20:18:16 <bl4ckb0ne> never thought id see `npm run startx` written anywhere
2022-12-13 20:18:36 <psykose> i have no idea what you're asking but to downgrade pipewire you have to build the older version yourself
2022-12-13 20:23:44 <chereskata> okey nice
2022-12-13 20:24:03 <chereskata> i thought alpine had something like openbsd ports
2022-12-13 20:24:33 <chereskata> so i can just change the APKBUILD to an older version
2022-12-13 20:24:40 <chereskata> and get the apk for installation
2022-12-13 20:24:54 <panekj> if you build it yourself, technically yes
2022-12-13 20:28:14 <psykose> you can do that in the aports repo by hand using git
2022-12-13 20:28:32 <chereskata> that was my idea
2022-12-13 20:28:56 <chereskata> git reset --hard ${comid} 
2022-12-13 20:29:03 <ikke> git checkout
2022-12-13 20:30:30 <chereskata> oh thanks, i used it just for branches
2022-12-13 20:46:52 <omni> I cache apks and usually have some of the old ones laying around
2022-12-13 20:48:55 <psykose> that also works
2022-12-13 21:01:01 <ikke> doing a patch upgrade of gitlab
2022-12-13 21:05:39 <psykose> did it get 10x faster? :)
2022-12-13 21:06:44 <ikke> *insert witty reply*
2022-12-13 21:07:38 <omni> how about a tenth as slow?
2022-12-13 21:07:55 <psykose> *insert laugh track 3*
2022-12-14 04:03:32 <sam_> psykose: https://discourse.gnome.org/t/multiple-fixes-for-gvariant-normalisation-issues-in-glib/12835
2022-12-14 04:03:52 <psykose> thanks
2022-12-14 04:04:07 <sam_> also unrelated https://gitlab.inria.fr/mpc/mpc/-/commit/e944aa454e60cbff8ab4e8c70dd974083398378f (and followup commits, i'd either drop 1.3.0 or just package 1.3.1_rc1, as it's 1.3.0+header fixes, i couldn't build gcc w/ 1.3.0)
2022-12-14 04:04:08 <sam_> np!
2022-12-14 04:05:10 <psykose> all the commits since 1.3.0 or?
2022-12-14 04:12:00 <sam_> so for me, 1.3.0 + https://gitlab.inria.fr/mpc/mpc/-/commit/e944aa454e60cbff8ab4e8c70dd974083398378f was enough to build gcc, but then they committed those extra patches since 1.3.0, so I decided to just deploy 1.3.1_rc1 to the masses and nothing broke, it all looked pretty harmless and i didn't want other stuff to be non-buildable
2022-12-14 04:12:07 <sam_> so yeah i'd do all the commits since 1.3.0
2022-12-14 04:12:18 <psykose> sure, testing all 6 for gcc
2022-12-14 04:32:11 <psykose> gcc passed
2022-12-14 04:32:12 <psykose> thanks
2022-12-14 04:48:24 <sam_> np
2022-12-14 08:15:26 <ncopa> those are the new kernel config knobs from 5.15 -> 6.1: https://tpaste.us/vyPL
2022-12-14 08:15:40 <ncopa> let me know if there is anything specific you'd like me to enable
2022-12-14 08:16:17 <ncopa> I have already enabled CONFIG_LRU_GEN
2022-12-14 08:25:21 <psykose> X86_AMD_PSTATE, X86_X32_ABI, ZSWAP, HID_NINTENDO, ANDROID_BINDER_IPC, SND_SOC_SOF_AMD_TOPLEVEL, SND_SOC_INTEL/AMD*
2022-12-14 08:28:53 <psykose> maybe reset_simple too
2022-12-14 09:54:27 <ncopa> why do you need X86_X32_ABI?
2022-12-14 09:54:39 <ncopa> we dont build anything for x32
2022-12-14 09:54:55 <ncopa> we dont even have toolchain for that
2022-12-14 11:24:16 <omni> I would like a(n additional) hardened config at some point
2022-12-14 12:10:28 <ikke> ncopa: is there a reason riscv64 has not been added to releases.yaml for edge?
2022-12-14 12:10:51 <ncopa> not really
2022-12-14 12:10:56 <ikke> ok
2022-12-14 12:11:04 <ncopa> i think we can add it
2022-12-14 12:11:26 <ncopa> ikke: do you have hardware that uses rtl8821ce-lts?
2022-12-14 12:11:38 <ncopa> it does not build with 6.1 kernel and i wonder if it is still needed
2022-12-14 12:11:41 <ikke> ncopa: no, I have not
2022-12-14 12:11:44 <ikke> It was for a user here
2022-12-14 12:11:45 <ncopa> ok
2022-12-14 12:12:13 <ikke> I don't mind dropping it, but I can check if there has been upstream changes
2022-12-14 12:14:21 <ncopa> that would be great. thanks!
2022-12-14 12:14:43 <ikke> There is a commit that adds 6.0 support
2022-12-14 12:15:07 <ikke> Would you be able to test if this commit works? 50c1b120b06a3b0805e23ca9a4dbd274d74bb305
2022-12-14 12:15:55 <ncopa> what git repo is that commit from?
2022-12-14 12:16:05 <ikke> https://github.com/tomaspinho/rtl8821ce
2022-12-14 12:16:17 <ncopa> ah, i already tested that
2022-12-14 12:16:21 <ikke> ok
2022-12-14 12:16:33 <ncopa> i updated it to latest commit. still failed. which is why i wonder if we can drop it
2022-12-14 12:16:46 <ncopa> i'm gonna drop the dahdi-linux-lts
2022-12-14 12:16:47 <ikke> We have a AKMS package for it as well
2022-12-14 12:16:56 <ikke> but it will probably have the same issue
2022-12-14 12:18:44 <ncopa> it will
2022-12-14 12:37:24 <ncopa> the listnewconfig for aarch64: https://tpaste.us/Keqb
2022-12-14 12:38:07 <ncopa> i think for now that i will simply use the defconfig for everything. most is off, and then we can enable on request
2022-12-14 12:39:02 <ncopa> CONFIG_ZSWAP=y CONFIG_DEVTMPFS_SAFE=y CONFIG_X86_AMD_PSTATE=y CONFIG_EFI_SECRET=m and some others, that applies to all arches
2022-12-14 12:45:00 <psykose> i was going to say 'the rockchip stuff' but indeed it's better if someone figures out everything needed for whatever new platform
2022-12-14 13:03:18 <chereskata> Hi
2022-12-14 13:03:26 <chereskata> is this the ssh fingerprint of gitlab.alpinelinux.org?
2022-12-14 13:03:28 <chereskata> The authenticity of host 'gitlab.alpinelinux.org (172.105.69.85)' can't be established.
2022-12-14 13:03:30 <chereskata> ED25519 key fingerprint is SHA256:vHA8oGFtHniAtj2LdIcp8OY3x6qhPqUEvNunwwNiahc.
2022-12-14 13:09:10 <psykose> yes
2022-12-14 13:09:17 <chereskata> thanks!
2022-12-14 13:10:05 <omni> ncopa: not saying you should keep dahdi but https://github.com/asterisk/dahdi-linux/pull/14
2022-12-14 13:21:21 <ncopa> huh. ok so people are still using it...
2022-12-14 13:21:34 <ncopa> I'll remove it and see what happens
2022-12-14 13:24:00 <chereskata> I have added all independently flagged ports to release-monitoring.org 
2022-12-14 13:24:38 <chereskata> I thought that this is a good moment to completely reset the flag database, because Anitya will flag them again 
2022-12-14 13:25:53 <chereskata> also i am curious where the source of pkgs.alpinelinux.org lies
2022-12-14 13:26:20 <chereskata> i would like to add a disclaimer that the user should add the package to Anitya instead of flagging it directly
2022-12-14 13:37:43 <Piraty> ncopa: yes, it's still used
2022-12-14 13:52:27 <omni> chereskata: I think it is https://gitlab.alpinelinux.org/alpine/infra/aports-turbo
2022-12-14 13:53:06 <chereskata> Thanks a lot
2022-12-14 14:02:37 <chereskata> ikke: is this database reset thing in pkgs.alpinelinux.org an easy task?
2022-12-14 14:04:29 <chereskata> (for flagged ports)
2022-12-14 17:13:20 <psykose> nmeum: ah, and i'm pretty sure the busybox testsuite now hangs again due to sigign sigpipe because of the aforementioned make 4.4 change for it. or maybe something else, but it's reproducible not even on the builders
2022-12-14 18:20:08 <omni> is this of interest? https://github.com/facebook/zstd/compare/v1.5.2...v1.5.2-kernel
2022-12-14 18:20:46 <psykose> what
2022-12-14 18:20:50 <psykose> 's interesting about it
2022-12-14 18:51:14 <ncopa> here comes linux-lts 6.1
2022-12-14 18:57:14 <psykose> did they tag it as lts already
2022-12-14 18:57:17 <psykose> i don't see it anywhere
2022-12-14 19:00:55 <ncopa> they didnt, but its usually the last release each year
2022-12-14 19:15:20 <ikke> ncopa: https://gitlab.alpinelinux.org/alpine/aports/-/issues/14457 can it be that it's strongswan that is adding `throw` routes on our ppc64le builder / ci host?
2022-12-14 20:11:12 <omni> psykose: idk, it just showed up in one of my feeds and I thought it might be 6.1 related
2022-12-14 20:18:19 <psykose> :)
2022-12-14 20:18:30 <psykose> in what sense? the kernel is still on 1.4.10 i think
2022-12-14 20:24:09 <ikke> static libs in alpine is like tides
2022-12-14 20:24:26 <ikke> and it's getting high tide
2022-12-14 20:25:04 <psykose> well this person just wants a static lib of everything to build random stuff it seems
2022-12-14 20:30:43 <PabloCorreaGomez[m]> BTW, thanks a lot ikke  for the merge in the qa-bot. Is the new version already deployed? Or are there any timing for it?
2022-12-14 20:30:53 <ikke> PabloCorreaGomez[m]: not yet, I have it on a todo list
2022-12-14 20:31:10 <PabloCorreaGomez[m]> Great, thanks a lot!
2022-12-14 20:31:11 <ikke> It has been due for a while, but this includes some changes in the deployment I need to take into account
2022-12-14 20:31:30 <ikke> I mean, other MRs
2022-12-14 20:31:50 <PabloCorreaGomez[m]> Perfect! Let me know if I can help with anything
2022-12-14 20:33:28 <ikke> It also means a change in the user that represents aports-qa-bot
2022-12-14 20:34:40 <ikke> The worst part is that I requested these changes :P
2022-12-14 20:36:02 <PabloCorreaGomez[m]> Jajajaja, good luck then :D
2022-12-14 20:36:14 <PabloCorreaGomez[m]> And in general, is there a plan for apk-tools v3 roll-out? Maybe Ariadne ? For the polkit frontend used by GNOME software, we would benefit from not hard-coding output to stdout. I saw apkv3 should have support for that, and backporting to v2 probably quite a lot of effort. So would like to avoid it if possible
2022-12-14 20:36:37 <ikke> That's for fabled to decide when he considers apkv3 to be ready
2022-12-14 20:45:49 <PabloCorreaGomez[m]> Ok, thanks! So fabled ? :D
2022-12-14 20:53:38 <q66> unfortunately apk-tools development has been inactive for a while now
2022-12-14 20:54:08 <q66> 
ACTION is developing a distro using apk3 and it's starting to become a bit of a problem

2022-12-14 20:54:39 <q66> but fabled said he's hoping to get back to it eventually
2022-12-14 20:56:19 <ikke> life happens
2022-12-14 20:58:54 <psykose> chimera-dpkg incoming
2022-12-14 21:10:27 <ncopa> ikke: could be strongswan that does it
2022-12-14 21:11:18 <ikke> So far that's the only server that I've seen this issue on though
2022-12-14 21:14:59 <q66> psykose: in the absolute worst case i'd fork apk-tools but for now i prefer to wait if upstream activity comes back
2022-12-14 21:19:56 <omni> should linux-headers also be upgraded to 6.1 then?
2022-12-14 21:20:10 <ncopa> yeah, i suppose so
2022-12-14 21:20:10 <omni> or just 6.0.13
2022-12-14 21:20:50 <ncopa> could be 6.1 but its not critical. not sure if it introduces any issues with musl headers
2022-12-14 21:21:15 <ncopa> linux-headers are only for userspace
2022-12-14 21:22:39 <psykose> omni: feel free to mr it
2022-12-14 21:47:00 <omni> ncopa: do you know more of linux-headers/revert-broken-uapi.patch ?
2022-12-14 21:52:19 <ncopa> i dont
2022-12-14 21:55:14 <psykose> (and i meant 6.1)
2022-12-14 21:56:32 <psykose> that one comes from ab4e5f1f008c40eb8c0894f81d304661e91f493a and iirc (i tried removing it) it breaks some header-level stuff paired with being included along with some musl headers
2022-12-14 22:21:44 <omni> psykose: !42340 happy? 🐈
2022-12-14 22:22:02 <omni> !42341
2022-12-14 22:22:51 <psykose> haha
2022-12-14 22:22:54 <psykose> yeah, no more mips
2022-12-14 22:23:35 <ikke> RIP
2022-12-14 22:24:29 <psykose> f
2022-12-14 22:25:08 <Reinhilde> mipped
2022-12-14 22:25:25 <Reinhilde> she no longer mips. one could say she mipped.
2022-12-14 22:52:34 <sam_> psykose: https://discourse.gnome.org/t/fixes-for-gdbusmenumodel-crashes-in-glib/12846 lol
2022-12-14 22:52:36 <sam_> (extra fixes)
2022-12-14 22:52:53 <sam_> i've sub'd to the discourse tag thingy now
2022-12-14 22:52:53 <psykose> already backported that
2022-12-14 22:52:56 <psykose> yeah same
2022-12-14 22:53:32 <psykose> surprised that it actually works
2022-12-14 22:53:39 <psykose> subscribing to 1st post in a tag
2022-12-14 22:53:57 <psykose> thought i'd never get any more emails ever again after the dropped the mls
2022-12-15 00:47:39 <sam_> me too
2022-12-15 00:47:55 <sam_> honestly very so often i want to make like
2022-12-15 00:47:57 <sam_> a packagers ml
2022-12-15 00:48:00 <sam_> just fo rshit like this 
2022-12-15 00:48:08 <sam_> "heads up X is fucked"
2022-12-15 00:51:05 <panekj> why ml tho
2022-12-15 00:51:44 <sam_> well, something else would work too
2022-12-15 00:51:48 <sam_> i don't mind what
2022-12-15 00:51:56 <sam_> just something where we can scream at each other with heads-upeseses
2022-12-15 01:00:01 <psykose> haven't seen a better system than an ML one yet for async notifications
2022-12-15 01:00:36 <psykose> if i get an email in any form i will definitely see it and it will stay there and i can do something with it, and it works
2022-12-15 01:00:46 <psykose> every other system does not do that, somehow
2022-12-15 01:00:55 <panekj> xmpp does that
2022-12-15 01:02:36 <psykose> kind of? it's like email but the spec is even worse and everything is even more broken
2022-12-15 01:03:10 <psykose> i didn't spend very long on xmpp but it was mostly just a broken mess
2022-12-15 01:03:17 <panekj> yes, it's so broken that every major company built upon XMPP
2022-12-15 01:03:22 <psykose> except gtalk that shit slapped
2022-12-15 01:03:56 <psykose> there is also the extremely sizable existing market part
2022-12-15 01:04:09 <psykose> almost everything lets you put in an email
2022-12-15 01:04:40 <psykose> and so i end up with email notifications for everything
2022-12-15 01:04:43 <psykose> quite hard to replace
2022-12-15 01:05:27 <panekj> to the big pile of hated broken thing, I should add another thing so they stick together and burn
2022-12-15 01:05:54 <panekj> btw. can someone ban that chinese spam on devel
2022-12-15 01:15:28 <sam_> psykose; absolutely
2022-12-15 01:15:30 <sam_> i'd be open to non-ml if there was a good suggestion but ml feels like a good fit
2022-12-15 01:16:11 <psykose> sam_: i wonder if anyone's tried this before
2022-12-15 01:16:18 <psykose> just some huge distro shouting place
2022-12-15 01:16:22 <psykose> i feel like it would have been
2022-12-15 01:24:23 <sam_> on the one hand i feel like it must have been tried
2022-12-15 01:25:26 <sam_> on the other hand i'm also not worried about popularity of it cause it works as long as u and me shout things at each other there 
2022-12-15 01:25:28 <sam_> i wonder who would know if it's been done
2022-12-15 01:40:54 <psykose> sounds good in theory but in practice i don't shout very much :p
2022-12-15 05:40:05 <fabled> PabloCorreaGomez[m], q66: I am having scheduled time for APK3 soon. Unfortunately there has been $-work emergencies, and life... So it got pushed more than desired. I'll probably try to do pending clean ups and major tickets. And do a pre/beta release. There's still work after those to do finish the code conversion for the new db format. But I wonder if we need to do 3.0 first, and leave part of the refactoring for 3.1.
2022-12-15 09:51:49 <PabloCorreaGomez[m]> That is great to hear! Life happens :) Could you ping here when you do a pre/beta release? So that maybe I can start looking at the polkit frontend changes?
2022-12-15 09:52:44 <PabloCorreaGomez[m]> Or do you think libapk's API will not change substantially with your refactoring, and I could start working on that already?
2022-12-15 13:26:37 <ikke> ncopa: https://gitlab.alpinelinux.org/alpine/aports/-/issues/14463 requests to update ca-certificates. https://hg.mozilla.org/mozilla-central/log/tip/security/nss/lib/ckfw/builtins/certdata.txt only shows NSS beta releases since the last upgrade. Does that matter?
2022-12-15 13:26:50 <ikke> (last one was at 2022-06-14)
2022-12-15 13:34:37 <ikke> https://gitlab.alpinelinux.org/alpine/ca-certificates/-/merge_requests/4
2022-12-15 13:40:26 <ncopa> ikke: no idea. probably not
2022-12-15 13:46:01 <ncopa> there is an nss-3.86 release
2022-12-15 13:46:22 <ncopa> https://bugzilla.mozilla.org/show_bug.cgi?id=1802996
2022-12-15 13:47:07 <ikke> All those commits refer that commit, but the commit messages somehow indicate it's still beta releases
2022-12-15 13:47:32 <ikke> Bug 1802996 - land NSS NSS_3_86_BETA1 UPGRADE_NSS_RELEASE, r=keeler
2022-12-15 14:18:00 <omni> ncopa: don't you also need to rebuild the community/*-lts aports against linux 6.1.0-r1?
2022-12-15 15:12:54 <minimal> ikke: that CA had a "Not After" of 8th Dec so it shouldn't actually be used by any (sane) software anyway
2022-12-15 15:14:16 <ikke> minimal: I have seen issues with expired root certs being present in the ca astore
2022-12-15 15:15:06 <minimal> ikke: in other words you mean software that didn't both to check the date validity of each part of the chain?
2022-12-15 15:15:11 <ncopa> omni: will do
2022-12-15 15:15:31 <ikke> minimal: no, validation using a chain that was expired and complaining
2022-12-15 15:16:36 <ikke> minimal: If I recall correctly, it was one of LEs earlier cross-signed certs
2022-12-15 15:17:13 <ikke> removing the expired cert caused it to use the still valid chain
2022-12-15 15:18:04 <minimal> ikke: ah I (vaguely) remember that, it was to do with some software having "selection" issues when multiple chains existed
2022-12-15 15:18:11 <ikke> yes
2022-12-15 15:18:26 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/issues/11607
2022-12-15 15:19:58 <minimal> ikke: more to the point, I assume the ca-certificates version upgrade now means that the CA (who's name I forget) who were recently stripped of "trusted" status by Mozilla, Google, etc will now be removed from Alpine's package...
2022-12-15 15:20:41 <ikke> TrustCor
2022-12-15 15:21:46 <ikke> It's still present, they marked it as distrust after
2022-12-15 15:22:19 <minimal> ah ok
2022-12-15 15:22:26 <ikke> "# For Server Distrust After: Wed Nov 30 00:00:00 2022"
2022-12-15 15:22:37 <ikke> Chrome will straight-out remove it apparently
2022-12-15 15:42:33 <minimal> ncopa: re the 6.1 kernel package upgrade, see https://gitlab.alpinelinux.org/alpine/mkinitfs/-/issues/29 which I raised several months ago
2022-12-15 15:52:03 <ncopa> right. i will have a look at that
2022-12-15 15:52:23 <ncopa> does anyone has such laptop can can help test?
2022-12-15 17:13:06 <psykose> ikke: all those -beta tags mean is that the changes were made then a beta was released. but then the actual nss version was released too
2022-12-15 17:13:25 <psykose> so these CA changes are in nss 3.86 stable, for instance
2022-12-15 17:14:02 <psykose> they were just made in development before the stable release, it's the same repo as nss after all
2022-12-15 18:20:29 <psykose> (i actually don't know why most of the changes to the files include the version they're going into in the commit message, it looks quite funny)
2022-12-15 19:29:10 <psykose> ikke: do you have a moment do give me an aarch64 container too
2022-12-15 19:52:52 <ikke> psykose: psykose-edge-aarch64.usa9.alpin.pow
2022-12-15 19:52:54 <ikke> psykose: psykose-edge-aarch64.usa9.alpin.pw
2022-12-15 19:53:45 <psykose> works, thanks
2022-12-15 19:59:55 <psykose> all set up
2022-12-15 20:01:32 <psykose> i'm also curious what even uses memory on that host, it's always half full with nothing building
2022-12-15 20:02:32 <ikke> the ci vms run on that host as well
2022-12-15 20:03:26 <ikke> using 64G of ram each
2022-12-15 20:03:34 <psykose> ah, right
2022-12-15 20:03:36 <psykose> reserved
2022-12-15 20:03:42 <psykose> makes sense
2022-12-15 21:10:19 <psykose> make[1]: /bin/sh: Argument list too long
2022-12-15 21:10:25 <psykose> can't even build libreoffice anymore these days
2022-12-15 21:10:30 <psykose> what on earth are they passing by argument
2022-12-15 21:11:34 <ikke> the entire dictionary?
2022-12-15 21:11:51 <elly> every object file in openoffice?
2022-12-15 21:11:58 <elly> or libreoffice rather
2022-12-15 21:12:06 <psykose> it's at the start of the build so something language related probably
2022-12-15 21:12:11 <psykose> still impressive as getconf ARG_MAX is like 131k
2022-12-15 21:12:37 <psykose> is there an easy way to get the failing cmdline there
2022-12-15 21:12:37 <psykose> hm
2022-12-15 21:26:33 <psykose> found it
2022-12-15 21:26:38 <psykose> it generates a filelist for something
2022-12-15 21:26:49 <psykose> contains every .idl
2022-12-15 21:26:59 <psykose> length: 499652
2022-12-15 21:28:17 <psykose> or rather it's the offapi_idl.filelist
2022-12-15 21:40:46 <sam_> i wish they didn't use their cheesy weird make wrapper thing
2022-12-15 21:40:48 <sam_> meson pls
2022-12-15 21:42:16 <psykose> no idea how to even fix that xd
2022-12-15 21:49:49 <omni> how to get that lua progressively rockin'?
2022-12-15 21:50:38 <elly> hm?
2022-12-15 21:51:44 <omni> !27474
2022-12-15 22:35:17 <elibrokeit> sam_: there was a proposal and they were shopping around for a GSOC effort, IIRC
2022-12-15 22:35:23 <elibrokeit> dunno what ever happened there.
2022-12-15 23:17:40 <panekj> ptrc: don't be *that* person, say how you fixed it
2022-12-15 23:20:39 <ptrc> panekj: fixed what?
2022-12-15 23:22:59 <panekj> aports-turbo
2022-12-15 23:25:22 <ptrc> i mean...the merge request is +3,-3 lines
2022-12-15 23:25:57 <panekj> i meant the comment, not what you are actually trying to fix
2022-12-15 23:26:22 <ptrc> ahhhh
2022-12-15 23:26:30 <ptrc> lmao right
2022-12-15 23:27:21 <psykose> confusion
2022-12-15 23:27:43 <panekj> deceive and evade
2022-12-15 23:29:21 <ptrc> i edited the comment :)
2022-12-15 23:30:01 <panekj> <3
2022-12-16 01:53:59 <sam_> elibrokeit: :(
2022-12-16 01:54:01 <sam_> psykose: i may have developed an rsi from tiff revbuilds
2022-12-16 01:54:03 <sam_> *rebuilds
2022-12-16 01:54:21 <psykose> tiff itself or was there some major revdeps needed recently
2022-12-16 01:54:47 <sam_> tiff itself changed soname today
2022-12-16 01:55:23 <psykose> o nice
2022-12-16 01:55:24 <psykose> new version
2022-12-16 01:55:41 <psykose> heh :) for me it takes like a few keystrokes
2022-12-16 01:55:54 <psykose> ap revdep tiff-dev | xargs apkgrel -ag
2022-12-16 01:56:03 <psykose> then some usual git
2022-12-16 01:56:37 <sam_> i finally wrote a script for it
2022-12-16 01:56:40 <sam_> i've been putting it off for ages
2022-12-16 01:56:51 <sam_> "eh there's not that many"
2022-12-16 01:57:26 <psykose> yeah it happens like all the time haha
2022-12-16 01:57:31 <psykose> i think you get your time back by the 3rd
2022-12-16 02:03:59 <psykose> i'm surprised by how many things have tiff support
2022-12-16 02:04:51 <psykose> i haven't seen a tiff file in over 6 years
2022-12-16 02:06:04 <sam_> 1. im gonna ask for a refund if not
2022-12-16 02:06:06 <sam_> 2. i know?!
2022-12-16 02:06:08 <sam_> is there like, some industry which is kept alive by tiffs or something
2022-12-16 02:06:19 <sam_> i know xrays or something come in some odd format
2022-12-16 02:07:17 <psykose> probably geo stuff
2022-12-16 02:07:32 <psykose> but that was specifically geotiff? shrug
2022-12-16 02:08:06 <psykose> and they just had to do the fuzzing nirvana on this rarely used library baked into everything :p
2022-12-16 02:08:17 <psykose> every 2 weeks there's a new cve
2022-12-16 02:08:42 <sam_> yeah you might be right actually, especially given the main guy who does it all is a geo dude
2022-12-16 02:08:43 <psykose> it almost gets as many cves as chromium somehow, impressive
2022-12-16 02:11:54 <minimal> psykose: I think often PDF files contain TIFF images, especially scanned documents
2022-12-16 02:11:59 <omegatron> sam_ : tiff is still widely used in DTP and DTP-alike stuff
2022-12-16 02:12:21 <omegatron> or, yes, medical context
2022-12-16 02:12:46 <psykose> surprised it's not an even older standard
2022-12-16 02:12:48 <omegatron> (and many others)
2022-12-16 02:13:29 <minimal> omegatron: aren't medical images usually something else? DICOM? (from memory)
2022-12-16 02:13:55 <minimal> yupe, DICOM
2022-12-16 02:14:15 <omegatron> DICOM is something like TIFF on steroids
2022-12-16 02:14:39 <omegatron> depends on what and how old your medical hard- or software is
2022-12-16 02:16:40 <omegatron> anyway, TIFF is still, besides EPS and PDF/PostScript used in companies/branches of advertising, printers, publishers, etc. (hence DTP)
2022-12-16 02:18:00 <sam_> thanks!
2022-12-16 02:18:02 <omegatron> (and often widely used, when it comes to 1-bit or 2-bit "color depth" images of large size)
2022-12-16 02:18:03 <elibrokeit> sam_: I thought gentoo already had revdep rebuild automation covered
2022-12-16 02:18:16 <sam_> elibrokeit: we do, but not if it's unexpected for something to change soname
2022-12-16 02:18:20 <sam_> you have to go around fixing deps
2022-12-16 02:19:13 <elibrokeit> no like, wasn't there some tool that scanned binaries with broken ldd loadage or something, and rebuilt them all
2022-12-16 02:19:23 <psykose> (insert movie voice) expect the unexpected!
2022-12-16 02:19:48 <elibrokeit> does it rely on tagging the soname in advance, then?
2022-12-16 02:20:02 <sam_> oh right yeah there is but we don't like relying on it
2022-12-16 02:20:29 <sam_> yeah, exactly, and then for cases like tiff, where it's considered kind of stable, you don't want to go around doing it proactively as people moan
2022-12-16 02:20:36 <sam_> https://projects.gentoo.org/qa/policy-guide/dependencies.html#proactive-use-of-slot-operators
2022-12-16 02:20:45 <sam_> after doing huge rounds of := deps for like the 10th time
2022-12-16 02:20:53 <sam_> i am possibly having a change of opinion 
2022-12-16 02:22:02 <sam_> for new libs, I do generally declare the soname because why not, so it's mostly an issue with older stuff
2022-12-16 02:22:39 <elibrokeit> what do they moan about
2022-12-16 02:25:59 <sam_> so like, take zlib - if i change the subslot now so it's accurate (and having a subslot is an indicator that consumers should use := when writing ebuilds, which is why it's useful to have) in preparation for An Event, you end up with every zlib consumer needing to be rebuilt even though nothing changed. but if i don't change it now and i wait until An Event happens, the flip side is I have to go around adding := into a million ebuilds at once 
2022-12-16 02:26:03 <sam_> the compromise would be add := even if something doesn't define a subslot yet
2022-12-16 02:26:05 <sam_> but the problem is a subslot doesnt strictly have to be soname or what you expect
2022-12-16 02:26:07 <sam_> like for qt, the subslot is for its private libraries
2022-12-16 02:26:11 <sam_> and for poppler, it's for one of its libraries, not all of them
2022-12-16 02:30:55 <elibrokeit> So you cannot add slots to the ebuild, while deferring changes to hundreds of revdeps until they need bumping anyway?
2022-12-16 02:33:59 <sam_> ah, so you can, but the problem is, some people proactively added := to their zlib deps when no subslot existed in the ebuild
2022-12-16 02:34:22 <sam_> at some point if i'm bored i think i'll just write some metadata mangling tool so people can skip the rebuilds if they want, and we just fix all the subslots in one go
2022-12-16 02:34:47 <elibrokeit> Ah, now I get it
2022-12-16 02:35:30 <elibrokeit> So "proactive use of slot operators" is a problem then?
2022-12-16 02:38:42 <omni> psykose: let's chat here instead of at gitlab
2022-12-16 02:38:52 <psykose> what's up
2022-12-16 02:39:56 <omni> if you are to add mbedtls 3.x, perhaps do it in community since it's not lts?
2022-12-16 02:43:14 <psykose> it is lts
2022-12-16 02:43:25 <omni> where does it say so?
2022-12-16 02:43:43 <psykose> ah misread
2022-12-16 02:43:58 <psykose> no rather, i think it would become lts, otherwise they won't have one anymore
2022-12-16 02:43:59 <psykose> not sure
2022-12-16 02:44:31 <psykose> but mostly the 2.28 branch goes out of support in 2 years so something would have to change, haha
2022-12-16 02:44:58 <psykose> nothing actually needs 3 so no reason to do it yet i guess
2022-12-16 02:45:02 <sam_> so much stuff is still broken with it sadly
2022-12-16 02:45:02 <omni> https://github.com/Mbed-TLS/mbedtls/releases/tag/v2.16.12 "This is the last release of the 2.16 long-time support branch. Users who want a long-time branch should move to mbedtls-2.28, which is backward-compatible and will be supported for at least 3 years."
2022-12-16 02:45:17 <psykose> most of the stuff is easy to fix
2022-12-16 02:45:30 <psykose> but then there's haxe and some other weird ones
2022-12-16 02:46:01 <sam_> elibrokeit: I think so, yeah. ideally nobody would do it, we'd add subslots to everything, then gradually add 
2022-12-16 02:46:02 <psykose> the majority of the porting was really just importing the compat header :p i'm sure the developers would actually fix it eventually
2022-12-16 02:46:03 <sam_> *:=
2022-12-16 02:46:37 <omni> that's why I went with 2.28.x over 3.1.0 at the time
2022-12-16 02:54:41 <omni> but what about 3.14 and 3.15? they're still at mbedtls 2.16.12
2022-12-16 02:56:31 <psykose> not sure
2022-12-16 02:58:21 <omni> at least I know I'm not up for the task of backporting fixes to 2.16.12
2022-12-16 03:22:05 <omni> I'm trying to make heads or tails of their plans wrt lts
2022-12-16 03:24:59 <omni> old posts before they moved their web, if anyone's interested:
2022-12-16 03:25:03 <omni> https://web.archive.org/web/20200809095126/https://tls.mbed.org/tech-updates/blog/announcing-lts-branch-mbedtls-2.16
2022-12-16 03:25:10 <omni> https://web.archive.org/web/20210728233748/https://tls.mbed.org/tech-updates/blog/working-towards-mbed-tls-3
2022-12-16 03:48:33 <EvTheFuture> I just upgraded my zoneminder VM and ran into some problems. The upgrade did not go well out of the box. The bigger issue though seem to be that zoneminder depend on the PHP class IntlDateFormatter which is not found. Does anyone know if this class is available as a package for Alpine and it's just some missing depend in zm?
2022-12-16 03:48:50 <EvTheFuture> I upgraded from Alpine 3.16 to 3.17
2022-12-16 04:01:03 <psykose> sam_: they also seem to have changed some stuff that gets rid of some TIFFFax* definitions
2022-12-16 04:01:16 <psykose> and you guessed it ancient garbage hylafax needs them
2022-12-16 04:01:47 <psykose> god this changelog
2022-12-16 04:01:59 <psykose> Major changes: 1 change: reformatted the source with clang-format
2022-12-16 04:02:00 <psykose> yeah okay
2022-12-16 04:12:58 <elibrokeit> honestly, that can be legitimately called a major change
2022-12-16 04:13:10 <elibrokeit> at least to anyone who has to work with the source code :p
2022-12-16 04:34:38 <psykose> i'm so confused why these symbols are gone
2022-12-16 04:34:53 <psykose> there's just a file with some const xyz abc=[]
2022-12-16 04:35:04 <psykose> and it's merely passed as a .o into a .so and all of them are gone
2022-12-16 04:35:54 <psykose> did i forget some obvious linking shit
2022-12-16 04:38:27 <sam_> yes
2022-12-16 04:38:45 <sam_> i think they defaulted to symbol versioning and set some visibliity stuff
2022-12-16 04:40:02 <psykose> i mean i manually set it as default and added -fvisibility=default everywhere but they still dissappear
2022-12-16 04:40:03 <sam_> or am i misunderstanding u and they're actually useful things which are gone
2022-12-16 04:40:03 <psykose> xd
2022-12-16 04:40:05 <sam_> libarchive freaked me out a few days ago until i realised ^ happened
2022-12-16 04:40:17 <psykose> they're gone, but i mean nothing uses that shit except hylafax
2022-12-16 04:40:19 <sam_> o
2022-12-16 04:40:25 <psykose> but they're not meant to be gone? nothing else uses this file
2022-12-16 04:40:31 <psykose> it's purely a generated .c with 3 symbols
2022-12-16 04:40:35 <psykose> not used anywhere in the tiff codebase
2022-12-16 04:40:38 <psykose> and linked into libtiff.so
2022-12-16 04:41:01 <psykose> tif_fax3sm.c
2022-12-16 04:42:19 <psykose> also they have a --enable-deprecated that when you pass, disables the deprecated things
2022-12-16 04:42:21 <psykose> fantastic
2022-12-16 04:42:57 <psykose> eyy
2022-12-16 04:43:02 <psykose> if i remove the version script
2022-12-16 04:43:03 <psykose> they appear!!
2022-12-16 04:43:04 <psykose> thanks
2022-12-16 04:43:12 <psykose> fantastic shit
2022-12-16 04:44:02 <psykose> ugh i see
2022-12-16 04:44:05 <psykose> that .map script
2022-12-16 04:44:08 <psykose> discards the symbols via linker
2022-12-16 04:44:08 <psykose> right
2022-12-16 04:44:11 <psykose> i forgot that was a thing
2022-12-16 04:44:22 <psykose> the one last piece of the linker puzzle
2022-12-16 04:44:27 <psykose> sam_: you're beautiful and i love you
2022-12-16 04:44:39 <sam_> <3
2022-12-16 04:53:35 <psykose> https://gitlab.com/libtiff/libtiff/-/issues/504
2022-12-16 04:53:41 <psykose> just in case you had hylafax to build
2022-12-16 06:18:57 <sam_> thank you!
2022-12-16 06:20:43 <ikke> EvTheFuture: php81-intl
2022-12-16 06:25:55 <psykose> accidentally typed rm * in my home folder
2022-12-16 06:25:57 <psykose> oops
2022-12-16 06:28:26 <ikke> oof
2022-12-16 06:28:28 <ikke> RIP
2022-12-16 06:28:51 <ikke> luckily that does not touch dot-files
2022-12-16 06:28:56 <psykose> aye
2022-12-16 06:37:42 <lotheac> no backups? :)
2022-12-16 06:43:22 <ikke> I typically don't have a lot of important files directly under $HOME/
2022-12-16 06:43:42 <ikke> no -r was passed, so it should only remove non-dot files in your home dir
2022-12-16 06:44:42 <psykose> at best i lost like an image
2022-12-16 06:44:45 <psykose> that i already got back
2022-12-16 06:44:52 <psykose> so nothin :)
2022-12-16 06:44:53 <psykose> was scary though
2022-12-16 06:45:00 <psykose> in other news
2022-12-16 06:45:17 <psykose> converting a random jpeg i had into jxl made it 20% smaller and added progressive decoding
2022-12-16 06:45:28 <psykose> passing it through avif made it 100% bigger and took 4x the cpu time
2022-12-16 06:45:31 <psykose> thanks chrome
2022-12-16 07:03:33 <ncopa> morning
2022-12-16 07:06:14 <ncopa> whats up with tiff?
2022-12-16 07:06:36 <ncopa> it seems it got updated with an ABI breaking change?
2022-12-16 07:08:02 <ncopa> oh, ok its a work in progress
2022-12-16 07:08:23 <ncopa> also, 3.17-stable seems to have broken deps in the repo now
2022-12-16 07:09:42 <ncopa> $ docker run --rm alpine:3.17 sh -c "apk upgrade -q -U -a && apk dot --errors" | tpaste
2022-12-16 07:09:42 <ncopa> https://tpaste.us/NOrP
2022-12-16 07:11:14 <ollieparanoid> I've also seen the libtiff errors. do you want a MR that bumps pkgrels of community packages?
2022-12-16 07:11:20 <ollieparanoid> apparently it has been done for main already, but not community
2022-12-16 07:11:53 <ncopa> psykose: do you want help with the tiff rebuilds?
2022-12-16 07:12:03 <psykose> they're already done
2022-12-16 07:12:24 <psykose> main just uploads first like every time
2022-12-16 07:12:34 <ollieparanoid> oh I see. thanks!
2022-12-16 07:12:36 <psykose> and then everything is broken for a few hours
2022-12-16 07:12:54 <ollieparanoid> right
2022-12-16 07:12:55 <psykose> i got trolled by a linker script for 2 hours but they are fine
2022-12-16 07:13:26 <ncopa> so we're just waiting for edge builders finish up community. good! thanks!
2022-12-16 07:13:26 <psykose> ncopa: xfce 4.18 is ready if you want to test it
2022-12-16 07:13:32 <ollieparanoid> heh. thanks for fixing it up so quickly. maybe one day I'll write a CI script that complains if rdeps need pkgrel bumps after soname change
2022-12-16 07:13:33 <ncopa> did you push it?
2022-12-16 07:13:41 <psykose> no, in !42374
2022-12-16 07:14:37 <psykose> ollieparanoid: i always merge it with the soname rebuilds, but for things with a hundred revdeps outside main.. :p
2022-12-16 07:14:41 <psykose> at least it's not as big as boost
2022-12-16 07:14:48 <psykose> or icu
2022-12-16 07:14:48 <ncopa> i'd like to test that, but I cannot build anything here righ now, til tiff is done
2022-12-16 07:14:58 <psykose> you could install the artifacts
2022-12-16 07:15:05 <ncopa> one by one?
2022-12-16 07:15:32 <psykose> i usually apk add *-r0.apk the folder or something
2022-12-16 07:15:45 <psykose> there is also mrtest
2022-12-16 07:15:58 <psykose> mrtest add -a 42374
2022-12-16 07:16:08 <ncopa> ok i might have a look at that later
2022-12-16 07:16:22 <ncopa> will try clean up the 3.17 dep tree today
2022-12-16 07:16:28 <psykose> what's wrong with it
2022-12-16 07:16:31 <ncopa> but first i need to go and get some breakfast
2022-12-16 07:16:39 <psykose> tasty breakfast
2022-12-16 07:16:40 <ncopa> broken deps: https://tpaste.us/NOrP
2022-12-16 07:17:06 <psykose> ah
2022-12-16 07:17:16 <ollieparanoid> psykose: I see, thanks for explaining
2022-12-16 07:21:35 <psykose> ncopa: fixed the lower ones
2022-12-16 07:21:44 <psykose> not sure why samba is circling itself or what happened with binsh there
2022-12-16 07:40:19 <psykose> ah, i guess it's technically circular? the -binsh symlinks depend on the parent since that's where it points to, but provide /bin/sh and the parent depends on one of them, so in theory it looks like a circle
2022-12-16 07:40:23 <psykose> it does work completely fine though
2022-12-16 07:40:33 <psykose> not sure if it's a real error at all but it looks annoying
2022-12-16 13:47:54 <omni> what am I doing wrong? when I have the 3.16-stable branch checked out and do `git pull` I get no later commit than 1b6934392c31c21a827afab531be55be43243962
2022-12-16 13:59:11 <ncopa> omni: and if you do git pull --rebase?
2022-12-16 14:00:02 <psykose> git status; git remote -v
2022-12-16 14:01:17 <omni> I moved forward with `git pull origin 3.16-stable`
2022-12-16 14:02:24 <omni> hmm..
2022-12-16 14:04:58 <omni> but `git status` gives "Your branch is ahead of 'omni/3.16-stable' by 89 commits." how did I get to this situation?
2022-12-16 14:05:04 <omni> and how do I get out?
2022-12-16 14:07:20 <ncopa> seems like your 3.16-stable branch follows omni/3.16-stable
2022-12-16 14:07:42 <omni> yes
2022-12-16 14:07:48 <omni> I'll figure it out
2022-12-16 14:11:13 <psykose> git pull upstream 3.16-stable
2022-12-16 14:11:29 <psykose> where upstream is the remote of alpine/aports
2022-12-16 14:11:51 <psykose> i bind 'git pull upstream' to 'g s' for that :)
2022-12-16 14:12:57 <omni> but for me "upstream" is "origin" I think?
2022-12-16 14:13:32 <omni> tried `git branch --track origin/3.16-stable` but perhaps I'm now only making things worse?
2022-12-16 14:13:41 <omni> at least it didn't help
2022-12-16 14:13:47 <psykose> none of that matters for the other command
2022-12-16 14:14:17 <psykose> in git remote you see something like `upstream	git@gitlab.alpinelinux.org:alpine/aports.git (fetch)`
2022-12-16 14:14:30 <psykose> whatever the name on the left is for alpine/aports
2022-12-16 14:14:52 <psykose> but to fix what you want to fix, git branch --set-upstream-to
2022-12-16 14:14:55 <psykose> same thing
2022-12-16 14:15:45 <omni> why is it like this for only thhis branch?
2022-12-16 14:16:19 <ncopa> someone has done a --set-upstream-to omni at some point
2022-12-16 14:20:47 <omni> now I've managed to make things worse!
2022-12-16 14:20:49 <omni> branch '3.16-stable' set up to track 'origin/master'
2022-12-16 14:20:50 <omni> :D
2022-12-16 14:24:08 <Habbie> applause
2022-12-16 15:02:17 <omni> you know what? fuck 3.16. that release is dead to me.
2022-12-16 15:02:23 <omni> (no, I'll sort it out later)
2022-12-16 16:10:09 <crzbear[m]> hehe git fun
2022-12-16 16:52:07 <omni> I solved it by removing any 3.16-stable branch, `git branch -D origin/3.16-stable`, checking out the latest commit of the 3.16-stable branch, `git checkout -b 3.16-stable` and then `git remote set-branches origin 3.16-stable`
2022-12-16 16:52:41 <omni> there's probably another way and I have no idea on how I managed to get to that state
2022-12-16 16:53:07 <omni> but now things are back to where they were
2022-12-16 16:53:10 <psykose> all you needed was --set-upstream-to
2022-12-16 16:53:49 <omni> I really really tried
2022-12-16 16:58:02 <omni> with `git branch --set-upstream-to origin`, origin/3.16-stable as well as with = instead of space
2022-12-16 16:58:13 <omni> I don't have in my buffer anymore the errors I got
2022-12-16 16:58:59 <omni> and I don't have anything similar to `git branch --set-upstream-to omni` in my history
2022-12-16 17:49:47 <omni> I'm sure it was due to an error on my part, I just don't know what!
2022-12-16 19:02:53 <EvTheFuture> ikke: I already had installed that one, but I found the problem. everything was upgraded to php81 but I had to manually install php81-fpm and add it with rc-update and remove php8-fpm
2022-12-16 19:44:31 <psykose> should be getting paid for this microsoft stuff for real
2022-12-18 03:37:44 <psykose> ikke: could you implement https://gitlab.alpinelinux.org/alpine/infra/aports-turbo/-/merge_requests/53 when you have the time so we can finally get rid of the prerelease notifications
2022-12-18 03:37:49 <psykose> (at least i'm finally pretty sure it works)
2022-12-18 18:35:59 <ddevault> SoC's are the actual worst
2022-12-18 18:46:45 <jhjacobs[m]> why so?
2022-12-18 18:47:04 <ddevault> no ability to debug boot issues
2022-12-18 18:47:13 <jhjacobs[m]> ahh :) fair enough
2022-12-18 18:47:16 <ddevault> been trying to get uart out of this raspberry pi for an hour
2022-12-18 18:47:49 <jhjacobs[m]> so far ive only been using raspberry's, so i'm eager to learn :)
2022-12-18 18:48:02 <jhjacobs[m]> my next SoC will be RISCV based
2022-12-18 21:12:53 <q66> ddevault: if you want early bootloader output on rpi uart, as far as I know you can't 
2022-12-18 21:13:12 <Habbie> (most other SoCs are better in that regard, i believe)
2022-12-18 21:13:45 <q66> it only shows up once kernel comes up, making it semi-useless
2022-12-18 23:46:44 <vkrishn> lbu leaves dir like /tmp/lbu-22281-1671406812-1736, after aborted operation, but does not clean it in next op, guess its intented ?
2022-12-18 23:47:26 <vkrishn> on small ram based install this can be a problem
2022-12-18 23:59:48 <vkrishn> maybe something like, https://tpaste.us/WQvR 
2022-12-19 12:30:26 <nmeum> psykose: re make 4.4 and SIGPIPE: it seems upstream changed SIGPIPE handling again https://git.savannah.gnu.org/cgit/make.git/commit/?id=92ab2e642d2c04b3dcb5a736ae6193680bfd5f74 I will try backporting this to see if it fixes busybox and m4 tests
2022-12-19 12:31:05 <nmeum> love how gnu make manages to introduce subtle breakage with each new release
2022-12-19 12:31:15 <psykose> :)
2022-12-19 12:31:18 <psykose> it's very fun indeed
2022-12-19 12:31:54 <psykose> but yeah nice they changed it back, should be ok then
2022-12-19 12:31:58 <psykose> thanks for keeping up with it
2022-12-19 12:59:00 <nmeum> is gitlab auto-converting emojis to images now? 😥
2022-12-19 12:59:08 <nmeum> https://gitlab.alpinelinux.org/-/emojis/2/relieved.png
2022-12-19 13:02:22 <psykose> hmm
2022-12-19 13:02:43 <psykose> it copies fine
2022-12-19 13:02:58 <psykose> unless the copy is overriden but the on page display is that
2022-12-19 13:03:34 <psykose> no, seems to be text on the page
2022-12-19 13:04:11 <psykose> they have it in a custom html tag so maybe it renders different for you
2022-12-19 13:31:08 <Piraty> anjan_: nice to see jami merged
2022-12-19 13:36:23 <Piraty> aww, it scales badly on pinephone
2022-12-19 14:39:44 <ddevault> is it safe to assume python rebuilds are underway on alpine
2022-12-19 14:39:48 <ddevault> alpine edge*
2022-12-19 14:41:51 <ddevault> no, they are not
2022-12-19 14:41:55 <minimal> ddevault: why? build.alpinelinux.org shows the Edge builders, apart from riscv64, are idle
2022-12-19 14:42:03 <ddevault> because my packages are broken
2022-12-19 14:42:09 <ddevault> python3 --version
2022-12-19 14:42:12 <ddevault> 3.11
2022-12-19 14:42:19 <ddevault> apk info -L py3-tqdm
2022-12-19 14:42:23 <ddevault> /usr/lib/python3.10
2022-12-19 14:43:14 <ddevault> ah, local cache issue
2022-12-19 14:43:52 <minimal> yes, py3-tqdm contents on Edge x86_64 show /usr/lib/python3.11
2022-12-19 14:44:11 <minimal> "apk -Ua upgrade" ?
2022-12-19 14:44:18 <ddevault> nah, I have a newer version in ~/packages for some reason
2022-12-19 19:07:47 <vkrishn> is it ok to get something like https://tpaste.us/wYda in dmesg  ?
2022-12-19 19:09:09 <vkrishn> though application works
2022-12-19 19:09:36 <ikke> looks like it crashed
2022-12-19 19:10:07 <vkrishn> no, I have ssh connected
2022-12-19 19:10:45 <vkrishn> and webserver response is ok
2022-12-19 19:11:28 <ikke> It could have been restarted
2022-12-19 19:14:48 <vkrishn> I haven't, noticed because it keeps scrolling in dmesg output
2022-12-19 19:15:45 <vkrishn> on slightly newer kernel 3.18.31, I get similar msg for git,sshfs
2022-12-19 19:16:10 <vkrishn> would try 4.x series next week
2022-12-19 19:17:45 <vkrishn> would try leaving ping on, see if response time increases
2022-12-19 19:20:01 <minimal> vkrishn: what has this to do with Alpine? 3.18.31 kernel? 4.x kernel?
2022-12-19 19:21:24 <vkrishn> alpine running on old kernel, sorry if its not alpine related
2022-12-19 19:21:43 <vkrishn> thought its wpa_supplicant problem
2022-12-19 19:22:50 <minimal> vkrishn: yes it is but running an old kernel like that means you have a franken-Alpine system which is not supported
2022-12-19 19:23:05 <vkrishn> ok
2022-12-19 19:53:10 <vkrishn> how are dtbs` put in releases/armv7/netboot-3.17.0/dtbs-lts/, I mean source ?
2022-12-19 21:10:26 <vkrishn> if its just copy from firmware/stockrom, I can push some, ones I am working on
2022-12-19 21:11:33 <vkrishn> EXYNOS-7870 board seems nice
2022-12-20 01:45:16 <vkrishn> hola ! \o/
2022-12-20 01:45:28 <vkrishn> minimal: all scary dmesg log gone, I gave franken aarch64 kernel
2022-12-20 01:45:40 <vkrishn> I think it just evolved, might soon be fit to wear alpine-tux
2022-12-20 03:14:13 <anjan_> yup! Piraty 
2022-12-20 07:36:36 <indy> hi all, i have small problem. i need simulate power outage in qemu/kvm/libvirt with database containing data from the past
2022-12-20 07:37:36 <indy> issue no 1. - there is script which is deleting old data (keep only 4 weeks back)
2022-12-20 07:38:24 <indy> issue no. 2 - setting libvirt xml config to set clock works fine, but only for poweroff/reboot
2022-12-20 07:39:50 <indy> so once i'll do 'virsh destroy testvm && virsh start testvm' time is back in 'now'
2022-12-20 07:42:19 <indy> after discussion on #virt channel with devels they wrote me, that changing <clock> in libvirt config doesn't prevent system to update time
2022-12-20 07:42:36 <indy> i have disabled ntp using 'setup-ntp none'
2022-12-20 07:42:55 <indy> and disabled also hwclock at boot
2022-12-20 07:54:56 <mercenary> just put a 'date 010123591984' in /etc/rc.local ?
2022-12-20 07:57:57 <indy> not working
2022-12-20 07:58:10 <indy> also /etc/loca.d/date.start
2022-12-20 07:58:25 <indy> i mean /etc/local.d 
2022-12-20 11:52:58 <chereskata> Hi, noticed that linux-lts in edge is v6.1, why so?
2022-12-20 11:53:33 <ikke> chereskata: In anticipation that 6.1 will be the next LTS kernel
2022-12-20 11:54:52 <chereskata> Good to know. Assumed it, but i was not sure. Been working on Anitya mappings for the linux kernel
2022-12-20 12:46:19 <chereskata> A lot of python packages are still mapped with py-* in Anitya
2022-12-20 12:46:22 <chereskata> https://release-monitoring.org/distro/Alpine/search/?pattern=py-
2022-12-20 12:46:37 <chereskata> They should be py3-* instead
2022-12-20 13:00:15 <chereskata> psykose: do you have interest in packaging librewolf?
2022-12-20 18:56:00 <sam_> browsers are generally (extremely) painful to package and keep up with
2022-12-20 19:11:28 <q66_> yet another reason to use webkit
2022-12-20 19:11:38 <q66_> easy to package, maintained by people who package
2022-12-20 22:02:57 <psykose> chereskata: there is no point to anitya map the kernel
2022-12-20 22:03:26 <psykose> and no there is no point to package librewolf either
2022-12-20 22:12:03 <pj[m]> you should perhaps close the issue then
2022-12-20 22:12:21 <pj[m]> https://gitlab.alpinelinux.org/alpine/aports/-/issues/12904
2022-12-20 22:16:09 <psykose> why, so someone can see me do that then start complaining that i said no to their preferred 40th browser?
2022-12-20 22:16:11 <psykose> :p
2022-12-20 22:19:56 <Ermine> Otherwise they would complain that Alpine devs ignore package requests.
2022-12-21 00:45:40 <anonymous> hi
2022-12-21 00:49:31 <bl4ckb0ne> psykose: could you pack ladybird plssssss
2022-12-21 00:49:39 <psykose> it's literally in the repo
2022-12-21 00:49:53 <psykose> and broken as fuck
2022-12-21 00:49:59 <anonymous> f
2022-12-21 00:50:00 <psykose> i am not sure why i accepted it
2022-12-21 00:50:48 <anonymous> find
2022-12-21 00:52:50 <bl4ckb0ne> yeet it
2022-12-21 00:53:58 <psykose> ptrc: wanna fix the above ^ :p
2022-12-21 03:17:25 <EvTheFuture> Does anyone have time/are able to verify the problem I have which prevent me from upgrading from 3.16 at the moment? It would be very helpful to know if this is a common problem or specific to my hardware/software combination: !14364
2022-12-21 03:17:36 <EvTheFuture> !14364
2022-12-21 03:18:04 <EvTheFuture> https://gitlab.alpinelinux.org/alpine/aports/-/issues/14364
2022-12-21 11:23:02 <ptrc> psykose: sure, i'll try :) it's currently way better than it was a while ago, but still behind and kinda broken sometimes
2022-12-21 11:23:33 <psykose> sometimes -> any js page /s
2022-12-21 11:24:04 <ptrc> not true, even alpine gitlab was working last time i checked :p
2022-12-21 11:29:07 <psykose> i don't believe you
2022-12-21 11:30:07 <psykose> https://img.ayaya.dev/rMjiDNuOOwif
2022-12-21 11:30:09 <psykose> why you lyin
2022-12-21 11:39:19 <ptrc> because.. it turns out my laptop has a magic build that i havent pushed? :3
2022-12-21 11:42:08 <psykose> typical ptrc keeping all the good software for herself
2022-12-21 11:46:22 <ikke> :D
2022-12-21 11:46:37 <Daanct12> doesn't we all do that? :P
2022-12-21 13:55:29 <chereskata> Does someone use multiple mirrors in /etc/apk/repositories?
2022-12-21 13:59:52 <psykose> like multiple of the same repository?
2022-12-21 14:00:00 <psykose> hmm
2022-12-21 14:00:02 <psykose> never tried
2022-12-21 15:34:52 <nindoja> what's the best way to debug abuild producing an apk in Alpine 3.17 that reports a package file format error? abuild fails at the  "Updating the tmp/x86_64 repository index.." stage and spits out the following error message: ERROR: runsafe-sbom-0.1.0-r0.apk: package file format error
2022-12-21 15:36:08 <nindoja> the APKBUILD file works for Alpine 3.14-3.16, and we're now moving to support 3.17. I tried to see if there were any format tweaks for packages in 3.17 but didn't see anything obvious that had changed
2022-12-21 15:37:37 <ikke> apk-tools did add some metadata validation iirc 
2022-12-21 15:38:49 <psykose> don't suppose you can share the apkbuild
2022-12-21 15:41:11 <nindoja> @psykose I can actually: https://pastebin.com/2t88JtGs
2022-12-21 15:41:38 <psykose> well, i do mean with the actual contents
2022-12-21 15:41:42 <psykose> but what that error means is uh
2022-12-21 15:41:42 <nindoja> well, one second. that's missing the shasums
2022-12-21 15:42:17 <psykose> i guess the most common cause is version based, but it's any of the metadata fields failing the spec
2022-12-21 15:43:11 <psykose> the commit that adds it is in .10 https://gitlab.alpinelinux.org/alpine/apk-tools/-/commit/a6126a6f2362ee47e91acb25083e1b112a57640a
2022-12-21 15:43:20 <psykose> and an example of abuild failing to set something https://gitlab.alpinelinux.org/alpine/abuild/-/issues/10083
2022-12-21 15:43:40 <psykose> so if you have a metadata field like depend = pc:gsettings-desktop-schemas>=40.alpha, that version there is invalid (by what apk accepts) so it fails
2022-12-21 15:43:43 <psykose> it could be any field
2022-12-21 15:43:45 <psykose> just means it's wrong
2022-12-21 15:43:59 <psykose> not very useful output but reading the pkginfo by hand should make it easy to see
2022-12-21 15:44:56 <nindoja> thanks for the info. I'll go poke around the metadata spec and see what we've got wrong
2022-12-21 15:46:09 <psykose> (you can easily peek the metadata with `tar xzfO file.apk .PKGINFO`)
2022-12-21 16:35:07 <chereskata> psykose: yes like having two mirrors for the alpine repos
2022-12-21 17:49:35 <pj[m]> chereskata: ask what your problem is
2022-12-21 17:50:11 <ikke> pj[m]: they left 
2022-12-21 17:50:26 <panekj> pffff
2022-12-21 22:37:46 <psykose> ptrc: the new one is the same
2022-12-22 10:44:31 <ikke> Hmm, on 3.17 where libcurl 7.86 is installed, installing curl will not upgrade libcurl, resulting in a broken curl. 
2022-12-22 10:45:23 <ikke> 'An unknown option was passed to libcurl' 
2022-12-22 10:54:21 <ncopa> we should have depends="libcurl=$pkgver" so we require the exact version
2022-12-22 12:24:59 <ikke> ncopa: !42559
2022-12-22 12:40:33 <ikke> !42559
2022-12-22 13:08:52 <ikke> flaky tests on s390x, fun
2022-12-22 23:57:45 <omni> another mirror that also is available as a tor hidden service (.onion) https://ftp.lysator.liu.se/pub/alpine/
2022-12-23 08:04:01 <pj[m]> ptrc: would you like to bump tailscale on edge? :>
2022-12-23 08:05:27 <psykose> what's stopping u
2022-12-23 08:07:21 <pj[m]> I don't have alpine
2022-12-23 08:07:32 <pj[m]> I'm also illiterate
2022-12-23 08:07:33 <pj[m]> and lazy
2022-12-23 08:08:06 <psykose> sounds like u got it real bad
2022-12-23 08:08:48 <pj[m]> it's a hard knock life
2022-12-23 09:58:50 <Piraty> since when did package query bot gain auto bump capabilities ?
2022-12-23 09:59:07 <ikke> package query bot?
2022-12-23 10:08:34 <psykose> what bump capabilities
2022-12-23 11:24:10 <clandmeter> is there a specific reason we do not have update-grub?
2022-12-23 11:24:45 <clandmeter> i guess we would copy the trigger to update-grub?
2022-12-23 11:25:30 <alpine-meetbot> clandmeter meant to say: i guess we could copy the trigger to update-grub?
2022-12-23 11:25:30 <clandmeter> s/would/could
2022-12-23 11:26:39 <ikke> I'm all for it
2022-12-23 11:26:57 <ikke> And then call update-grub from the trigger
2022-12-23 11:28:01 <clandmeter> yep
2022-12-23 11:28:12 <clandmeter> but maybe there is another way of which i dont know.
2022-12-23 11:28:23 <ikke> apk fix grub
2022-12-23 11:28:38 <clandmeter> apk fix linux-xxx :)
2022-12-23 11:28:42 <ikke> right
2022-12-23 11:28:59 <ikke> It's hard to discover :P
2022-12-23 11:29:01 <clandmeter> but i find that a bit weird to just update the config :)
2022-12-23 12:00:08 <skarnet> apk dwim
2022-12-23 12:15:37 <donoban> I just had a font problem opening a pdf with xpdf, and fixed creating a symlink to '/usr/share/fonts/Type1' on '/usr/share/ghostcript/fonts' , maybe there is some default conf error or xpdf is not well built?
2022-12-23 12:16:57 <psykose> how did you know to do that
2022-12-23 12:17:14 <donoban> I straced xpdf and look the paths it failed to read
2022-12-23 12:17:24 <psykose> hmm
2022-12-23 12:17:25 <donoban> I can paste it if it helps
2022-12-23 12:18:56 <psykose> i guess ghostscript-fonts needs the symlink
2022-12-23 12:23:33 <donoban> I don't know if more apps rely on that path, if it's only xpdf it could also patched on /etc/xpdfrc
2022-12-23 12:24:20 <psykose> shrug
2022-12-23 12:24:21 <psykose> added anyway
2022-12-23 12:24:44 <donoban> wow :)
2022-12-23 12:46:21 <clandmeter> ikke: raising the cma size on boot removes the cma errors on the new arm server
2022-12-23 12:46:31 <clandmeter> raised to 64M
2022-12-23 12:46:48 <ikke> aha ok
2022-12-23 12:48:01 <clandmeter> not sure if we actually need it, and if it brings any performance differences.
2022-12-23 12:51:15 <psykose> probably no negative effect other than losing a few MB for other uses given that it's just reserved memory for a specific use
2022-12-23 12:55:53 <donoban> psykose: I tested your fix but if fails, it just creates an empty dir
2022-12-23 12:56:02 <psykose> i am high sorry
2022-12-23 12:56:07 <psykose> mistyped :)
2022-12-23 12:56:40 <donoban> hehe I supposed that was something wrong when saw it but I tested before saying anything
2022-12-23 12:57:30 <psykose> the upgrade will fail because it's not a dir anymore, but was fast so i left it like that
2022-12-23 12:58:42 <donoban> ok, I will remove and reinstall
2022-12-23 12:58:52 <psykose> a mere fix also works
2022-12-23 12:58:56 <psykose> i think
2022-12-23 12:58:58 <psykose> probably
2022-12-23 12:59:19 <donoban> I tried it after removing my symlink but nothing appeared
2022-12-23 13:00:29 <psykose> because i'm stupid x2 and forgot the pkgrel
2022-12-23 13:00:29 <ikke> pkgrel was not bumped
2022-12-23 13:00:32 <psykose> sorry, just tired
2022-12-23 13:00:52 <psykose> this time for real
2022-12-23 13:01:57 <donoban> a pkgrel waring on githooks would be pretty helpful
2022-12-23 13:02:16 <donoban> what % of aports commits don't bump it?
2022-12-23 13:02:37 <psykose> upgrades
2022-12-23 13:02:43 <psykose> moves
2022-12-23 13:02:44 <psykose> secfixes
2022-12-23 13:02:50 <psykose> random style shit with no rebuild needed
2022-12-23 13:03:13 <donoban> well if version is changed the pkgrel would be ignored or check that it goes to 0 again
2022-12-23 13:03:35 <ikke> It doesn't happen that often that it's forgotten
2022-12-23 13:03:41 <ikke> and often realized soon enough
2022-12-23 13:04:44 <psykose> yep
2022-12-23 13:04:59 <donoban> well, just a suggestion, could it be interactive?
2022-12-23 13:05:14 <psykose> you can write whatever githooks you want for yourself :)
2022-12-23 13:05:14 <ikke> Users can install local hooks
2022-12-23 13:05:17 <psykose> they're always personal
2022-12-23 13:05:22 <ikke> remote hooks cannot be interactive
2022-12-23 13:05:25 <donoban> "Pkgver&pkgrel unchanged, package won't be rebuilt, do you want to continue? y/n'
2022-12-23 13:05:32 <psykose> then you can even share them and people can opt in
2022-12-23 13:05:40 <psykose> i really don't like project automatic githooks myself
2022-12-23 13:05:49 <psykose> but i do use them
2022-12-23 13:05:55 <ikke> I do have a local hook that warns if a commit message is incorrect (does not start with <repo>/<package>
2022-12-23 13:06:26 <ikke> Mostly because I use automaticaly variables which can sometimes fail
2022-12-23 13:06:47 <ikke> $pkg is automatically set to <repo>/<pkgname> if I cd into a package
2022-12-23 13:06:53 <ikke> so I can do git commit -m "$pkg: ..."
2022-12-23 13:07:28 <donoban> hehe I see
2022-12-23 13:07:41 <donoban> well it seems fine now, upgraded failed but apk fix works
2022-12-23 13:07:51 <psykose> yep
2022-12-23 13:07:59 <psykose> hopefully nobody caught that in the 30 minutes :p
2022-12-23 13:08:06 <psykose> annoyingish apk issue
2022-12-23 17:09:42 <minimal> clandmeter: yes having a update-grub script invoked by a trigger would make sense, however such a script typically only runs grub-mkconfig - what I've always found strange is that there is no script (not sure if a trigger would be deemed "risky") to run grub-install to update the MBR (for BIOS) or the EFI file(s) for UEFI - it seems that once Alpine initially installs Grub those are never updated, even when the Grub version changes (which 
2022-12-23 17:09:42 <minimal> potentially introduces either new functionality or bug fixes to those))
2022-12-23 17:10:39 <minimal> so if you installed Alpine back when it used Grub 2.04 and since updated to Grub 2.06 your MBR or EFI files would still be from Grub 2.04
2022-12-23 19:41:18 <ikke> meson 1.0.0 released
2022-12-23 19:41:27 <ikke> I bet psykose is already on it
2022-12-23 23:53:46 <hiohoihoi> hi
2022-12-23 23:53:48 <hiohoihoi> any way to install virtualbox in alpine linux?
2022-12-24 00:21:55 <panekj> no
2022-12-24 00:22:08 <panekj> but they left!
2022-12-24 00:22:16 <panekj> ffs
2022-12-24 00:22:48 <minimal> panekj: the asked in the alpine-linux channel at same time and got a reply there
2022-12-24 00:23:05 <panekj> great, another reposter
2022-12-24 00:23:34 <minimal> not a reposter, a simultaneous multi-channel poster
2022-12-24 00:24:07 <panekj> that would be if they posted same message at the same time
2022-12-24 00:24:14 <panekj> in both channels
2022-12-24 00:24:56 <minimal> panekj: within 50 seconds in the 2 channels
2022-12-24 00:25:07 <minimal> so effectively simultaneously ;-)
2022-12-24 05:01:58 <psykose> ikke: heh
2022-12-24 05:02:06 <psykose> tbf it's just stabilisation :) barely any changes in it
2022-12-24 08:35:08 <ncopa> psykose: I think 8f7d80c6d4cc9ad1cc2fd2f79d51becd7c5a1abc may break things
2022-12-24 08:35:28 <psykose> what specifically
2022-12-24 08:35:39 <ncopa> linux-rpi
2022-12-24 08:35:49 <ncopa> _flavors=
2022-12-24 08:36:32 <ncopa> reason is how the APKBUILDs are parsed during build time
2022-12-24 08:36:58 <ncopa> to avoid fork for every APKBUILD, which would be very slow, i just source them
2022-12-24 08:37:36 <ncopa> for i in */APKBUILD; do . $i ...;done
2022-12-24 08:38:11 <ikke> ncopa: would running it in a subshell help? Or does that add forking again?
2022-12-24 08:38:40 <ncopa> running in subshell for each APKBUILD would solve it, but it means fork
2022-12-24 08:39:00 <ncopa> i mean, yeah, it would work, it would just take a minute to parse the tree instead of a second
2022-12-24 08:39:03 <psykose> added that one back
2022-12-24 08:39:47 <ikke> https://gitlab.alpinelinux.org/alpine/go/-/tree/master/apkbuild :P
2022-12-24 08:40:14 <ncopa> so, anything that does _var="$_var something" should start with an empty assignment
2022-12-24 08:41:53 <psykose> does that apply to zfs-lts/rpi that doesn't do that but still has for _extra_flavors and set only sometimes
2022-12-24 09:37:15 <ptrc> hm, might be a weird question but.. why source multiple apkbuilds in the same shell?
2022-12-24 09:38:17 <ptrc> because i thought about that and discarded it as "i don't think anyone actually does this"
2022-12-24 09:42:36 <psykose> it's faster than a shell for each
2022-12-24 09:44:53 <ptrc> yeah, but
2022-12-24 09:45:23 <ikke> to prevent the overhead of forking 10k times
2022-12-24 09:45:40 <ptrc> if someone's sourcing all the apkbuilds, surely they don't capture variables like _flavors for every single one of them
2022-12-24 09:45:52 <psykose> hmm
2022-12-24 09:46:03 <psykose> yeah actually i wonder when it's relevant
2022-12-24 09:46:05 <ikke> ptrc: after the APKBUILD is sourced, the variable is read
2022-12-24 09:46:28 <psykose> for actual building the subshelling would be fine
2022-12-24 09:46:31 <ikke> for F in */*/APKBUILD; . $F; echo $pkgname $_flavor; done
2022-12-24 09:46:48 <psykose> i don't think that is used for anything
2022-12-24 09:46:55 <ikke> ncopa has scripts for automating updating kernel packages
2022-12-24 09:47:27 <ikke> buildrepo does the same, but there those variables not matter
2022-12-24 09:47:35 <ikke> or I must say, lua-aports
2022-12-24 09:48:45 <psykose> reasoning about shells is certainly impossible if the compatibility is set at "someone has some scripts somewhere that might do anything with this file"
2022-12-24 09:49:24 <psykose> it could have merely been kernel_flavors="" and a real variable
2022-12-24 09:50:27 <psykose> insofar as builds i don't think i've broken anything so far, but for people that could have been doing anything with the files i've probably caused a hundred breaking changes
2022-12-24 09:50:30 <psykose> this isn't really well defined
2022-12-24 18:09:20 <panekj> what was the config option to enable funky faster git?
2022-12-24 18:09:43 <ikke> does not ring a bell
2022-12-24 18:10:00 <ikke> or do you mean protocol v2?
2022-12-24 18:10:13 <panekj> I believe protocol v2 should be default?
2022-12-24 18:10:22 <panekj> I'll just check logs
2022-12-24 18:11:08 <ikke> yes, we had to enable it on the server, which we did
2022-12-24 18:11:32 <panekj> 2022-10-02 12:06:45 <ikke> thanks btw for mentioning it, was not aware that was missing / required for protocol v2
2022-12-24 18:11:36 <panekj> yep
2022-12-24 18:12:34 <panekj> maybe I misremember something, but I recall psykose or ptrc talking something about having enabling something in client side config as well
2022-12-24 18:13:17 <psykose> don't think so
2022-12-24 18:13:24 <psykose> for that it's default to v2 for a while
2022-12-24 18:13:53 <panekj> anyway, it pull'd surprisingly fast
2022-12-24 18:13:55 <panekj> > Updating eba8f9b543..28fc4a5a58
2022-12-24 18:14:51 <psykose> i also use controlmaster but that's unrelated
2022-12-24 18:15:16 <panekj> git supports controlmaster?
2022-12-24 18:15:42 <panekj> (server)
2022-12-24 18:15:58 <psykose> it's a clientside feature
2022-12-24 18:16:25 <panekj> I meant that does it allow lingering connections
2022-12-24 18:16:31 <psykose> sure
2022-12-24 18:16:35 <panekj> since that's what CM does
2022-12-24 18:19:04 <panekj> eh, makedepends="go" could implicitly enable options="net"
2022-12-24 18:22:29 <psykose> when you get the message, hang up the phone
2022-12-24 18:23:18 <panekj> sorry, can't hear you, I'm installing factorio
2022-12-24 18:23:48 <psykose> IR3 just released : )
2022-12-24 18:23:55 <psykose> might have to play through it next week
2022-12-24 18:26:08 <panekj> imagine having time for playing games
2022-12-24 18:26:30 <ikke> IR3?
2022-12-24 18:26:41 <panekj> industrial revolution 3
2022-12-24 18:26:55 <panekj> factorio mod
2022-12-24 18:27:22 <psykose> the person making it really cares about graphics so it looks beautiful :p
2022-12-24 18:27:23 <ikke> ah
2022-12-24 18:27:34 <bl4ckb0ne> factiorio runs on alpine?
2022-12-24 18:27:42 <panekj> tailscale.I_Acknowledge_This_API_Is_Unstable = true
2022-12-24 18:27:43 <psykose> didn't actually try
2022-12-24 18:27:51 <psykose> i use flatpak steam for all the games
2022-12-24 18:28:08 <psykose> works the same as flatpak steam in any distro
2022-12-24 18:28:30 <psykose> but sure lets give it a go
2022-12-24 18:28:43 <ikke> panekj: what API?
2022-12-24 18:28:57 <panekj> containerboot
2022-12-24 18:29:14 <panekj> Curious that it *seems* to work in rootbld
2022-12-24 18:29:56 <panekj> oh well, just as I said that: "FAIL    tailscale.com/cmd/containerboot 600.108s "
2022-12-24 18:33:35 <psykose> bl4ckb0ne: nope, segfaults with gcompat :)
2022-12-24 18:35:42 <bl4ckb0ne> lovely
2022-12-24 18:35:48 <bl4ckb0ne> in a dep?
2022-12-24 18:36:00 <bl4ckb0ne> do thry hace a gog version?
2022-12-24 18:36:32 <psykose> they have a download off their site lol
2022-12-24 18:36:34 <psykose> and no deps
2022-12-24 18:37:23 <panekj> what the hell is a headless factorio
2022-12-24 18:37:38 <bl4ckb0ne> might be worth bullyibg the for a musl cpmpatible build
2022-12-24 18:38:00 <bl4ckb0ne> fwiw i got the itch.io version celeste working on alpine
2022-12-24 18:38:11 <bl4ckb0ne> had to recompile some deps
2022-12-24 18:39:58 <psykose> it's factorio but headless
2022-12-24 18:40:10 <psykose> with cli options
2022-12-24 18:40:16 <psykose> so you can host a multiplayer server somewhered
2022-12-24 18:55:24 <panekj> mhm
2022-12-24 18:55:48 <panekj> factorio.alpinelinux.org
2022-12-24 20:46:09 <Ermine> PureTryOut: hi, are you the author of pipewire-launcher script? If yes, is there a reason why it resides in libexec?
2022-12-24 20:51:39 <panekj> probably because it's not supposed to be started by user?
2022-12-24 20:53:07 <PureTryOut> exactly that
2022-12-24 20:53:32 <panekj> as everything that resides in libexec
2022-12-24 20:54:43 <panekj> I should probably scream at people who put stuff in PATH instead of libexec for some stuff
2022-12-24 20:55:39 <panekj> 
ACTION screams at jirutka

2022-12-24 20:56:29 <Ermine> panekj: thank you, bu I know the purpose of libexec
2022-12-24 20:57:20 <Ermine> But, anyway, pipewire-launcher was mentioned as a way to start pipewire on a wiki. Even before I modified that page.
2022-12-24 20:59:27 <panekj> ask someone who put it there then
2022-12-24 21:20:32 <psykose> something something fhs makes no sense
2022-12-25 07:16:09 <NaTaSahaa> 
ACTION https://t.ly/orB1

2022-12-25 16:39:32 <ikke> upgraded gitlab to 15.5
2022-12-26 04:06:54 <psykose> poggers
2022-12-26 04:49:00 <psykose> 6.2-rc1 finally has the best toggle ever
2022-12-26 04:49:05 <psykose> Allow legacy TIOCSTI usage (LEGACY_TIOCSTI) [Y/n/?] (NEW)
2022-12-26 04:53:41 <psykose> which finally gets rid of https://github.com/ncopa/su-exec/issues/23 and friends
2022-12-26 05:26:16 <ptrc> aw, i wish it was a togglable feature though; TIOCSTI is useful sometimes
2022-12-26 05:28:34 <psykose> it is togglable
2022-12-26 05:28:58 <psykose> and it is never a useful feature
2022-12-26 09:35:58 <Ermine> psykose: yaaaaaay
2022-12-26 09:40:14 <Ermine> psykose: there were some attempts to do something like this, but Alan Cox NACKed them
2022-12-26 09:40:30 <psykose> acked now i guess : )
2022-12-26 09:41:39 <Ermine> Alan apparently wasn't involved this time :D
2022-12-26 10:06:24 <Ermine> Nice Christmas present!
2022-12-26 15:43:22 <pj[m]> ikke: clandmeter: did the issues with docker-abuild CI got resolved?
2022-12-26 15:45:12 <ikke> pj[m]: I don't think so. I suppose the issue was a lack of multi-arch image builders, right?
2022-12-26 15:45:59 <pj[m]> I don't remember the exact issue, there was something about it being different than rest of CI in gitlab.a.o
2022-12-26 15:46:30 <pj[m]> thank you algitbot, you are active that one time when actually not needed
2022-12-26 15:46:56 <ikke> yes
2022-12-26 15:50:31 <pj[m]> I'm currently going through pain of "exec /bin/sh: exec format error" trying to build my own images once again
2022-12-26 15:51:03 <pj[m]> because docker is fun stuff: The requested image's platform (linux/s390x) does not match the detected host platform (linux/amd64) and no specific platform was requested
2022-12-26 15:52:33 <ikke> I mean, that entire project can use an update. Latest release it mentions is 3.11
2022-12-26 15:53:01 <ikke> pj[m]: do you get that when you just pull the image?
2022-12-26 15:53:36 <pj[m]> Yes, I know lol I have more recent images on docker hub
2022-12-26 15:53:38 <pj[m]> https://hub.docker.com/repository/docker/panekjdev/docker-abuild
2022-12-26 15:54:21 <pj[m]> I remember someone here was also using dabuild so I would like to eventually update them
2022-12-26 15:54:30 <pj[m]> (and dabuild too)
2022-12-26 15:55:02 <pj[m]> ikke: no, I'm trying to update my images on docker hub
2022-12-26 15:55:11 <pj[m]> I think qemu is borked
2022-12-26 15:55:39 <ikke> You need to be very carefull when working with docker images for multiple arches
2022-12-26 15:56:58 <ikke> But I also have 'native' hosts for each arch to build the images on
2022-12-26 15:57:32 <ikke> docker-dabuild builds all images on a single host with qemu-binfmt
2022-12-26 15:57:56 <ikke> I guess carlo needs to decide what he wants with the project
2022-12-26 15:57:58 <pj> I know
2022-12-26 15:58:18 <ikke> sorry, didn't mean to imply you didn't know
2022-12-26 15:58:26 <pj> no problem :)
2022-12-26 15:58:35 <pj> I'm doing this all from my macos
2022-12-26 15:58:53 <pj> it's a weird experience
2022-12-26 15:59:06 <ikke> An extra layer of weirdness
2022-12-26 16:00:33 <pj> I never learned qemu stuff, so I'm always lost when it has issues
2022-12-26 16:00:53 <ikke> I know barely enough
2022-12-26 16:01:16 <pj> for docker, I just run "docker run --privileged --rm tonistiigi/binfmt --install all" and hope it works
2022-12-26 16:02:56 <pj> is docker socket available in CI?
2022-12-26 16:03:24 <ikke> There are runners that have it, yes, but they are not generally available
2022-12-26 16:03:51 <pj> so I can't just steal some runner if I make project? :P
2022-12-26 16:03:52 <ikke> everything under the alpine/infra/docker namespace can use it
2022-12-26 16:04:30 <pj> seems like I fixed qemu stuff, the magic command saved me again
2022-12-26 16:04:34 <ikke> heh
2022-12-26 16:04:46 <ikke> I am working on our docker ci infra atm, so maybe in the near future
2022-12-26 16:05:41 <pj> I haven't used Alpine or anything Alpine-specific in months so it's a bit hard to come back and fix stuff
2022-12-26 16:06:08 <ikke> I'd rather not make these docker image builders general available, but I'm working on setting up project runners that I can enable on a per-project (user) basis
2022-12-26 16:06:09 <pj> I'll be trying to maintain aports from mac, will see how that goes
2022-12-26 16:06:26 <pj> and I need dabuild for that
2022-12-26 16:07:23 <ikke> https://gitlab.alpinelinux.org/alpine/infra/docker/exec/-/tree/master/docker-image
2022-12-26 16:07:58 <pj> there is internal registry now?
2022-12-26 16:08:28 <ikke> yes
2022-12-26 16:08:35 <pj> \o/
2022-12-26 16:08:45 <ikke> Still experimenting with it
2022-12-26 16:08:55 <pj> docker-cli
2022-12-26 16:08:57 <pj> boooo
2022-12-26 16:08:59 <pj> bad
2022-12-26 16:09:04 <ikke> lol
2022-12-26 16:09:22 <pj> real devs just curl the api :P
2022-12-26 16:09:42 <ikke> real devs just talk to raw sockets :P
2022-12-26 16:10:45 <pj> that entrypoint stuff looks suspiciously similar to dabuild scripts
2022-12-26 16:11:13 <ikke> Not supprising
2022-12-26 16:11:29 <pj> alpinelinux/docker-abuild                                              edge-s390x          e4bd12fb485e   3 minutes ago    248MB
2022-12-26 16:12:38 <ikke> The idea is to create reusable building blocks that many projects need
2022-12-26 16:12:56 <ikke> instead of repeating it everywhere
2022-12-26 16:13:09 <pj> it sounds cool on paper
2022-12-26 16:13:21 <pj> but if it's going to be like with build images then that sucks
2022-12-26 16:14:32 <pj> like https://gitlab.alpinelinux.org/alpine/infra/docker/package-builder/
2022-12-26 16:15:06 <pj> I remember I had to click 3 different projects because each referenced docker image of another
2022-12-26 16:19:32 <ikke> You mean https://gitlab.alpinelinux.org/alpine/infra/docker/alpine-gitlab-ci?
2022-12-26 16:20:40 <pj> maybe, I tried to setup builders and tried to understand whole configuration
2022-12-26 16:21:06 <ikke> That builds the image that is used for aports CI
2022-12-26 16:21:16 <ikke> Which uses build-base
2022-12-26 16:21:42 <pj> that would be good to know months ago :P
2022-12-26 16:22:05 <pj> I don't remember specifics
2022-12-26 16:26:04 <pj> Unsupported arch "arm64" detected.
2022-12-26 16:26:06 <pj> Expected one of: x86|x86_64|aarch64|armhf|armv7
2022-12-26 16:26:12 <pj> lol
2022-12-26 16:26:52 <ikke> alpines architecture naming is slightly different from what docker uses :)
2022-12-26 16:27:04 <ikke> https://gitlab.alpinelinux.org/alpine/infra/docker/exec/-/blob/master/docker-image/entrypoint#L22
2022-12-26 16:27:48 <pj> that's actually uname -m output on mac
2022-12-26 16:28:44 <pj> I long for the days where I don't have to think about x86_64 = amd64, armv8 = aarch64 = arm64, etc.
2022-12-26 16:29:04 <ikke> The nice thing about standards is that there are so many of them
2022-12-26 16:30:21 <pj> pick your poison type of situation
2022-12-26 21:08:18 <pj> is there a reason why abuild-fetch is written in C (still)?
2022-12-26 21:08:40 <psykose> what else would it be written in
2022-12-26 21:09:18 <pj> /bin/sh ?
2022-12-26 21:09:32 <pj> like other abuild scripts ?
2022-12-26 21:09:42 <ikke> It has file locking in case of shared distfiles
2022-12-26 21:10:39 <psykose> flock is usable in the shell
2022-12-26 21:10:59 <psykose> though with less control i guess
2022-12-26 21:11:20 <pj> bb doesn't have flock
2022-12-26 21:11:36 <pj> and there is no cmd:flock
2022-12-26 21:11:42 <ikke> https://pkgs.alpinelinux.org/package/edge/main/x86_64/flock
2022-12-26 21:11:57 <psykose> what?
2022-12-26 21:12:07 <psykose> https://img.ayaya.dev/NyCTdE4vTyrs
2022-12-26 21:12:43 <psykose> cmd:flock also works to add the util-linux one
2022-12-26 21:13:04 <ikke> There are 4 other tools in c in abuild btw
2022-12-26 21:13:30 <ikke> tar, sudo, rmtemp, gzsplit
2022-12-26 21:14:12 <pj[m]> 
ACTION uploaded an image: (22KiB) < https://matrix.org/_matrix/media/v3/download/matrix.org/upikuQDSoKXbGPIztBhlEJUw/image.png >

2022-12-26 21:14:14 <pj[m]> psykose
2022-12-26 21:14:43 <pj[m]> turns out getting disconnected from server and having apk on macOS might be confusing
2022-12-26 21:14:55 <ikke> heh :D
2022-12-26 21:15:21 <pj> ikke: I know, I just have slight issues with afetch right now since it has hardcoded path
2022-12-26 21:15:50 <psykose> what has a hardcoded path
2022-12-26 21:16:08 <pj> the thing I'm complaining about
2022-12-26 21:16:20 <ikke> /var/cache/distfiles?
2022-12-26 21:16:38 <pj> yes
2022-12-26 21:16:44 <pj> char *destdir = "/var/cache/distfiles";
2022-12-26 21:16:58 <psykose> not really an issue
2022-12-26 21:17:06 <pj> not really asked you
2022-12-26 21:17:24 <psykose> is it a requirement to have permanent skill issues if one is from poland
2022-12-26 21:17:37 <ikke> please don't make it personal
2022-12-26 21:17:44 <pj> is it a requirement to be massive arsehole when not asked?
2022-12-26 21:18:26 <psykose> i'd generally ask you that
2022-12-26 21:18:32 <q66_> it is an issue to be from poland :^)
2022-12-26 21:19:01 <pj> I'm an arsehole always but at least I'm aware of that :')
2022-12-26 21:54:47 <psykose> https://img.ayaya.dev/VqwZH8oEGPDQ this is enough to make it dynamic but abuild.conf also has SRCDEST
2022-12-26 21:54:56 <psykose> i guess it could also be named SRCDEST but that is a teeny name
2022-12-26 21:55:18 <psykose> steal and submit whatever works for you i suppose
2022-12-26 22:31:09 <pj> I just changed it to my own hardcoded value lol
2022-12-26 22:31:51 <psykose> :)
2022-12-26 22:32:01 <psykose> to be fair it does look like a slight oversight since srcdest is exactly for that
2022-12-26 22:32:58 <psykose> yeah
2022-12-26 22:33:04 <psykose> it's been there for over 12 years
2022-12-26 22:33:09 <psykose> and the old sh code fetched into srcdest
2022-12-26 22:33:17 <psykose> then it got replaced with abuild-fetch and got hardcoded
2022-12-26 22:33:30 <psykose> so the exact patch above but with srcdest would match the old behaviour
2022-12-26 22:33:53 <psykose> yoink it and enjoy a free fix :p
2022-12-26 23:02:58 <anton> Hi.
2022-12-26 23:02:58 <anton> I'm trying to make an app, using Hyper-V socket but it turned out there's no hv_sock module out of the box.
2022-12-26 23:02:58 <anton> Could you recommend a manual how to build a kernel module?
2022-12-26 23:06:31 <pj[m]> curl google.com?q=how%20to%20build%20kernel%20module
2022-12-26 23:08:22 <pj[m]> or you can read APKBUILD for hvtools
2022-12-26 23:09:30 <pj> and wiki article how to make alpine package
2022-12-26 23:09:33 <minimal> anton: there is the hvtools package, however it is missing some stuff (I have patches I need to get around to finishing up and submitting a MR for them)
2022-12-26 23:16:48 <minimal> anton: the likely problem is that neither linux-lts nor linux-virt Alpine kernel packages have CONFIG_HYPERV_VSOCKET set
2022-12-26 23:17:14 <minimal> so no hv_socket either compiled into the kernel nor built as loadable module
2022-12-27 00:20:10 <pj> tbf, if bb has flock, I would just rewrite it in sh
2022-12-27 00:20:25 <pj> and look into removing rest of C code
2022-12-27 00:20:47 <pj> since abuild is sh (and it's not going away anytime soon) it would be better to have less dependencies
2022-12-27 02:12:25 <elibrokeit> why not just use curl/wget directly 
2022-12-27 02:21:11 <q66_> it was already said
2022-12-27 02:21:14 <q66_> because of file locking
2022-12-27 08:21:24 <clandmeter> ikke: we still need file locking?
2022-12-27 08:21:46 <ikke> clandmeter: yes
2022-12-27 08:21:52 <ikke> aarch64/armhf/armv7
2022-12-27 08:22:21 <clandmeter> hmm
2022-12-27 08:22:27 <clandmeter> i thought it was specifically used for x86
2022-12-27 08:22:34 <clandmeter> where we connected the servers
2022-12-27 08:22:54 <ikke> Those share the same distfiles dir as well
2022-12-27 08:23:14 <ikke> ncopa even had to fix an issue because it was not properly working in that setup
2022-12-27 08:23:40 <psykose> it's needed locally too just by being able to run multiple abuilds
2022-12-27 08:26:24 <ncopa> morning
2022-12-27 08:26:43 <ncopa> it is still not 100% robust
2022-12-27 12:06:15 <vkrishn> instead of relying on flock, something like this might just work, https://tpaste.us/NOlP
2022-12-27 12:16:13 <psykose> not quite sure what that is meant to do or why it is any better than one flock
2022-12-27 12:23:45 <vkrishn> slightly better, https://tpaste.us/x54o
2022-12-27 12:24:27 <vkrishn> psykose: it can be used to download file in distfiles, where multiple download is initiated for same file
2022-12-27 12:25:24 <psykose> that already works?
2022-12-27 12:26:25 <vkrishn> well someone said ^"tbf, if bb has flock, I would just rewrite it in sh"
2022-12-27 12:27:44 <psykose> and that would use flock and do it in 2 lines
2022-12-27 12:27:48 <psykose> i'm not sure what this is doing
2022-12-27 12:27:55 <psykose> downloading them at the same time for both fetches?
2022-12-27 12:28:02 <ikke> Just retrying until succesfull it appears to me
2022-12-27 12:29:06 <ikke> vkrishn: note that the builders are typically triggered at the same time, so they probably keep clobbing each-other each attempt
2022-12-27 12:29:19 <ikke> clobbering8
2022-12-27 12:29:29 <vkrishn> ok
2022-12-27 12:29:45 <psykose> currently all that happens is one starts the fetch and the others stay paused on the lock
2022-12-27 12:29:56 <psykose> once the fetch is done they all just use the same one
2022-12-27 12:29:57 <psykose> in theory anyway
2022-12-27 12:30:10 <ikke> yes, and at the moment, that works well
2022-12-27 12:30:16 <vkrishn> someone was to eager to use "^why not just use curl/wget directly"
2022-12-27 12:30:25 <psykose> you can
2022-12-27 12:30:34 <psykose> though hopefully not with whatever you wrote there
2022-12-27 12:30:47 <psykose> use flock instead, i promise it makes more sense :)
2022-12-27 12:31:45 <ikke> note that abuild-fetch uses record locks as well to work over NFS
2022-12-27 12:33:02 <ikke> Oh, it switched to lockf
2022-12-27 12:33:06 <ikke> https://gitlab.alpinelinux.org/alpine/abuild/-/commit/3da770ce3557975660162793e00239ac91a7af68
2022-12-27 12:34:00 <psykose> s6-setlock also works over nfs since it uses fcntl
2022-12-27 12:34:49 <psykose> lockf is the same interface yeah
2022-12-27 12:34:52 <psykose> built on fcntl(2)
2022-12-27 12:36:17 <psykose> well, unspecified i guess, not sure who does what in practice
2022-12-27 12:36:49 <psykose> musl is just a fcntl wrapper
2022-12-27 12:38:19 <vkrishn> if someone does try to convert abuild-fetch to sh script, this might be useful
2022-12-27 12:59:00 <q66_> i implemented locking in cbuild just yesterday and i ended up using a global flock over the whole sources dir
2022-12-27 12:59:08 <q66_> per-file locking just kinda doesn't work
2022-12-27 12:59:13 <q66_> without complicating things too much
2022-12-27 12:59:33 <ikke> for our use-case, it does work
2022-12-27 12:59:39 <ikke> at least, I haven't seen any issues lately
2022-12-27 13:00:33 <q66_> in my case the primary problem is that templates may share sources, and cbuild stores a by-sha256 cache so that it can use the same source file for both templates (via a hardlink) and avoid downloading it again if not needed etc
2022-12-27 13:01:50 <q66_> the accesses to this cache as well as the fact that an unrelated template may be downloading technically the same source file at the same time would introduce a bunch of complexity and having to think about what the critical sections should be without introducing races
2022-12-27 13:02:08 <q66_> so just locking the whole dir is robust and in practice doesn't really make anything worse
2022-12-27 13:03:01 <q66_> also afaict unlinking a lockfile always introduces potential races
2022-12-27 13:04:22 <q66_> as long as you use an advisory mechanism that works on file descriptors associated with specific inodes, there will always be a chance of race if you remove that file anywhere
2022-12-27 13:06:40 <q66_> (that means reliably doing a per-file lock without races means having a million lockfiles that linger around)
2022-12-27 14:10:36 <ncopa> btw, what what the problem with abuild-fetch in the first place?
2022-12-27 14:10:47 <ncopa> you can set download directory with -d
2022-12-27 14:20:09 <ikke> pj was anoyed that it was written in c
2022-12-27 14:23:47 <ikke> and abuild even passes SRCDEST to abuild-fetch
2022-12-27 14:34:32 <ncopa> pj: this was because we needed lock files that worked over nfs. See the commit message. https://gitlab.alpinelinux.org/alpine/abuild/-/commit/92186b70ca9c520fc726e0885aac633aadace655
2022-12-27 14:36:29 <q66_> ncopa: what does that mean
2022-12-27 14:36:43 <q66_> the point of emulating it with a fcntl *is* to make it work over nfs
2022-12-27 14:37:08 <ikke> q66_: it was later changed to use fcntl
2022-12-27 14:37:14 <ikke> or lockf
2022-12-27 14:37:15 <q66_> because traditional flock mechanism does not
2022-12-27 14:37:34 <q66_> see local_lock in https://man7.org/linux/man-pages/man5/nfs.5.html
2022-12-27 14:37:36 <vkrishn> its positive if someone does write that, is a sign of build new distros on top of AL
2022-12-27 14:37:48 <vkrishn> and they might not use nfs
2022-12-27 14:38:17 <q66_> (this has been the case since kernel 2.6.12)
2022-12-27 14:38:46 <q66_> ikke: the point is that flock should work over nfs just fine
2022-12-27 14:38:48 <vkrishn> spin off distros
2022-12-27 14:38:51 <q66_> as long as you have kernel >= 2.6.12
2022-12-27 14:40:39 <q66_> (and as long as the nfs is not mounted with local_lock=flock, which makes the locks apply only locally)
2022-12-27 14:43:01 <q66_> oh, i think i understand
2022-12-27 14:43:34 <q66_> the problem happens if you run abuild on a machine where the mount is a local filesystem, and then on another machine where it's an nfs mount
2022-12-27 14:43:55 <ikke> That was the exact situation
2022-12-27 14:43:56 <q66_> the nfs mount will use a posix lock, while the serverside abuild will use an flock
2022-12-27 14:44:05 <vkrishn> I would really like to see gparted/sysrescuecd style dists based on AL
2022-12-27 14:44:22 <vkrishn> or desktops
2022-12-27 14:44:50 <q66_> that said i think your locking is still racey due to the lockfile unlink
2022-12-27 14:45:10 <vkrishn> this would moved some queryies to their site
2022-12-27 14:54:11 <q66_> funnily in python the only difference to fix this is to change fcntl.flock call to fcntl.lockf
2022-12-27 14:54:25 <q66_> the flags are otherwise the same
2022-12-27 15:26:26 <pj> ikke: annoyed might be an overstatement, I was curious at most
2022-12-27 15:27:05 <ikke> Ok, misunderstood you then
2022-12-27 15:27:18 <ikke> communicating over text is hard sometimes
2022-12-27 15:27:21 <pj> true
2022-12-27 15:27:41 <pj> although You are right on point about me preferring sh :P
2022-12-27 15:28:28 <skarnet> don't tell it to anyone, and you didn't hear it from me, but sh is written in C
2022-12-27 15:28:41 <q66_> rewrite sh in sh
2022-12-27 15:28:55 <skarnet> that's my first goal for 2023
2022-12-27 15:29:11 <q66_> can't spell shell without hell
2022-12-27 15:29:17 <pj> rewrite sh to be glibc only
2022-12-27 15:29:28 <pj> no sh, no problem
2022-12-27 15:30:12 <pj> anyone actually uses nfs these days...?
2022-12-27 15:30:42 <skarnet> unfortunately yes, I know at least one setting that does
2022-12-27 15:30:56 <skarnet> and doing otherwise would be difficult
2022-12-27 22:14:36 <omni> can't have shit without sh
2022-12-28 00:29:47 <omni> as in shit's essential
2022-12-28 01:11:38 <q66_> it goes both ways
2022-12-28 04:56:00 <Guest1662> hey, wondering why package libexecinfo-dev isn't available in 3.17 / edge repos
2022-12-28 04:58:23 <ptrc> spiralw: libexecinfo is utterly broken on musl
2022-12-28 04:58:35 <ptrc> https://gitlab.alpinelinux.org/alpine/aports/-/commit/50795a14dee639ce2dcc836e2b2baca9bad4a1b1
2022-12-28 05:00:03 <spiralw> ah i see
2022-12-28 06:48:48 <spiralw> does musl not support LD_DEBUG?
2022-12-28 06:52:44 <psykose> no
2022-12-28 06:52:58 <pj[m]> LD_DEBUG is glibc shit if I'm not mistaken
2022-12-28 19:07:19 <vkrishn> ikke: somewhere near your keyboard ?
2022-12-28 19:49:34 <ikke> I'm here now
2022-12-28 19:54:06 <ikke> PabloCorreaGomez[m]: fyi, I'm still working on deploying the new aports-qa-bot, but it requires some more changes
2022-12-28 19:56:33 <ikke> Interesting: https://manned.org/pkg/alpine-3.17
2022-12-28 23:18:37 <fossy> hi, i have been investigating an issue when PXE booting from EFI on RPI4 where network connection could not be established. based upon https://github.com/pftf/RPi4/issues/137, i recompiled linux-lts with CONFIG_MDIO_BCM_UNIMAC=y (instead of CONFIG_MDIO_BCM_UNIMAC=m), and as suspected, this fixed the issue. however i am uncertain as to whether this is the -correct- solution.
2022-12-28 23:19:36 <fossy> should I; set CONFIG_MDIO_BCM_UNIMAC=y, adjust netboot initramfs to include mdio_bcm_unimac and dependent modules (i think these can then be loaded using modules= kernel cmdline), or something else?
2022-12-28 23:21:13 <fossy> (also, what is the correct way to modify dotconfigs to ensure dependencies? i used make menuconfig, which worked fine, but produces an enormous diff, which makes me think it is not the correct way...)
2022-12-28 23:32:13 <minimal> fossy: you'd have to include the ntp client in the initramfs and modify the initramfs' init to run the NTP client
2022-12-28 23:32:53 <minimal> fossy: oops, replying to your alpine-linux port in wrong channel ;-)
2022-12-28 23:44:02 <fossy> minimal: :P, no worries
2022-12-29 08:19:36 <vkrishn_> https://manned.org/pkg/alpine-3.17 - busybox-extras missing 
2022-12-29 08:20:19 <ikke> There is no busybox-extras-docs package
2022-12-29 08:20:30 <ikke> busybox-extras-doc*
2022-12-29 08:27:38 <vkrishn> Currently available applets include - should show httpd, unless thats removed in 3.17
2022-12-29 08:49:07 <ncopa> morning
2022-12-29 08:49:23 <ncopa> where can i find the linter that encourages this: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41600/diffs?commit_id=16b8a34728016559dc152a4f84721dd033cd50bc#note_280848
2022-12-29 08:50:30 <ikke> https://gitlab.alpinelinux.org/Leo/atools
2022-12-29 08:50:34 <ncopa> there is no guarantees that  _variables are unset on invocation
2022-12-29 08:51:40 <ncopa> so `_flavors=` or `unset _flavors`  is required when extending it with: `_flavors="$_flavors .."`
2022-12-29 08:53:51 <ikke> The goal for the linter is to remove the default variables that are not used, not necessarily custom variables
2022-12-29 08:54:36 <ncopa> https://gitlab.alpinelinux.org/Leo/atools/-/blob/master/apkbuild-lint#L255
2022-12-29 08:55:03 <mercenary> _flavors='' would keep linter happy, and result predictable
2022-12-29 08:55:48 <ncopa> or unset _flavors
2022-12-29 08:56:23 <ncopa> but i think the linter warning is a bit dangerous
2022-12-29 08:56:39 <ncopa> we should probably ignore _variables
2022-12-29 08:56:41 <mercenary> unset is probably better in this case
2022-12-29 08:57:08 <ncopa> its perfectly valid to set it to empty: _flavors=
2022-12-29 08:57:24 <ncopa> the linter is wrong in this case
2022-12-29 08:57:28 <ptrc> maybe there should also be a linter rule to check if variable is being appended to without first emptying/unsetting it?
2022-12-29 08:59:00 <ptrc> also, the linter can check if variable is being appended to (AL36)
2022-12-29 08:59:15 <ptrc> so someone would just have to add the same check to AL5
2022-12-29 09:52:41 <ncopa> sounds like a good idea
2022-12-29 10:26:32 <ikke> Or just being assigned at all
2022-12-29 10:26:51 <ikke> _foo=; ..; _foo=bar should not give a warning
2022-12-29 10:27:29 <ikke> (maybe that's what you meant ptrc)
2022-12-29 11:03:45 <mercenary> alias yes_really=''; yes_really _foo=; ...; _foo=bar; passes the linter
2022-12-29 11:08:24 <ptrc> ikke: imo your example should still be covered by the linter rule
2022-12-29 11:08:58 <ikke> Hmm, true
2022-12-29 11:09:05 <ikke> No need to first set it empty
2022-12-29 11:09:38 <mercenary> what if the example was _foo=; ...; if quux; then _foo=bar; fi
2022-12-29 11:11:19 <ptrc> ...hm, fair
2022-12-29 12:29:44 <PureTryOut> Is there a package that can be installed to get Alpine artwork on the machine like a logo, preferably in svg?
2022-12-29 12:54:05 <psykose> haven't seen one
2022-12-29 12:57:20 <ikke> same
2022-12-29 13:45:25 <raspbeguy> hello
2022-12-29 13:45:52 <raspbeguy> I just discovered that prometheus has a LTS version
2022-12-29 13:46:38 <raspbeguy> how can it be managed package side ?
2022-12-29 13:46:59 <raspbeguy> could we add a prometheus-lts package?
2022-12-29 13:47:02 <pj[m]> prometheus-lts?
2022-12-29 13:47:05 <pj[m]> yes
2022-12-29 13:48:22 <pj[m]> but since it's written in Go, it will be a subject to how long Go is supported in Alpine
2022-12-29 13:48:50 <raspbeguy> hum I just checked on repology and no other distro seem to have that
2022-12-29 13:50:17 <psykose> what would the point of packaging it be
2022-12-29 13:50:20 <raspbeguy> well, you need go only at build time
2022-12-29 13:50:57 <raspbeguy> psykose, for people wishing to get only bugfixes and secfixes
2022-12-29 13:51:10 <ptrc> raspbeguy: but the go runtime is embedded in the executable, so if there are any vulnerabilities, the executable itself would likely not get rebuilt if it's not supported anymore by alpine
2022-12-29 13:51:31 <raspbeguy> ptrc, good point
2022-12-29 13:58:21 <pj[m]> it's more of a, you can't move it to main because go is not in main
2022-12-29 13:58:31 <pj[m]> so what's the point in having lts for 6 months
2022-12-29 14:03:13 <raspbeguy> prometheus LTS version is less than 6 months
2022-12-29 14:03:39 <raspbeguy> oh I was wrong
2022-12-29 14:03:43 <raspbeguy> https://prometheus.io/docs/introduction/release-cycle/
2022-12-29 14:04:06 <vkrishn> will there to much of work/complications if .dbg pkgs are moved to separate dir (not in aports) but on uploads area
2022-12-29 14:32:16 <pj[m]> <raspbeguy> "prometheus LTS version is less..." <- that would be weird, but if we are not using lts version currently, we probably should
2022-12-29 14:33:50 <raspbeguy> pj[m], I was wrong on that sentence
2022-12-29 14:46:07 <pj[m]> yes, I read your messages, my point still applies :)
2022-12-29 14:47:18 <pj[m]> ddevault: any reason why prometheus package is going with CR versions instead of LTS?
2022-12-29 14:56:52 <raspbeguy> in my opinion we should keep both lts and non-lts
2022-12-29 14:58:56 <pj[m]> non-lts wouldn't get any updates
2022-12-29 14:59:08 <pj[m]> also it would be silly to keep it
2022-12-29 15:00:43 <raspbeguy> I don't get it
2022-12-29 15:01:23 <raspbeguy> would you mind explining It a bit further to me please?
2022-12-29 15:01:55 <pj[m]> I have to explain alpine's policy on updating packages in community?
2022-12-29 15:02:42 <pj[m]> packages are supposed to have security/bug fixes only
2022-12-29 15:03:19 <ikke> non-breaking updates are fine as well
2022-12-29 15:03:52 <raspbeguy> pj[m], you mean on stable release
2022-12-29 15:04:27 <raspbeguy> but that shouldn't be the case on edge
2022-12-29 15:04:50 <pj[m]> what is on edge ends up being in stable release
2022-12-29 15:05:16 <raspbeguy> yes, but that's why having both makes sense to me
2022-12-29 15:06:19 <pj[m]> how
2022-12-29 15:06:44 <ikke> We have both nodejs and nodejs-current
2022-12-29 15:06:52 <pj[m]> at some point a non-lts release would end up in stable release and would be a subject to alpine policy on updating packages
2022-12-29 15:06:54 <ikke> one lts, one latest release
2022-12-29 15:08:18 <pj[m]> nodejs has longer current release support
2022-12-29 15:08:42 <ikke> I guess it's non-existing for prometheus current?
2022-12-29 15:08:44 <pj[m]> (also quite weird)
2022-12-29 15:09:00 <pj[m]> because nodejs current releases turn into lts
2022-12-29 15:09:13 <pj[m]> (some)
2022-12-29 15:09:45 <pj[m]> what could be done is if prometheus non-lts stayed in testing
2022-12-29 15:09:57 <pj[m]> that way it doesn't end up in stable release while still being on edge
2022-12-29 15:10:09 <raspbeguy> I was thinking about somthing like this
2022-12-29 15:10:10 <pj[m]> (but that's abusing testing channel)
2022-12-29 15:10:20 <raspbeguy> but I don't know that's testing purpose
2022-12-29 15:10:35 <raspbeguy> but I don't think that's testing purpose
2022-12-29 15:10:56 <pj[m]> it is, in fact, not
2022-12-29 15:11:22 <raspbeguy> should we get a new edge-specific repo?
2022-12-29 15:11:41 <pj[m]> for one package?
2022-12-29 15:11:47 <raspbeguy> of course not
2022-12-29 15:12:11 <raspbeguy> but I guess prometheus isn't the only one project that has lts versions
2022-12-29 15:12:20 <raspbeguy> node for example
2022-12-29 15:12:34 <pj[m]> I'm not going to say anything as I'm not the one tirelessly maintaining whole aports and builders
2022-12-29 15:12:51 <pj[m]> node-lts is in stable
2022-12-29 15:13:11 <pj[m]> node-current is also fine because each nodejs release is supported for at least 6 months
2022-12-29 15:13:58 <raspbeguy> and node-lts?
2022-12-29 15:14:06 <pj[m]> longer
2022-12-29 15:15:02 <psykose> 3 years
2022-12-29 15:15:33 <pj> nodejs (lts) is in main and nodejs-current is in community
2022-12-29 15:15:36 <pj> as it should be
2022-12-29 15:15:52 <raspbeguy> I see
2022-12-29 15:16:38 <pj> community gets a bit of leeway because it's maintained mostly by community
2022-12-29 15:17:03 <pj> so not every package is properly vetted to adhere to alpine policies
2022-12-29 15:21:25 <raspbeguy> so every packages which major versions have a support version shorter than 6 months shouldn't be included in alpine repos?
2022-12-29 15:21:46 <raspbeguy> support period*
2022-12-29 15:21:52 <pj> technically
2022-12-29 15:22:04 <raspbeguy> ok
2022-12-29 15:26:56 <raspbeguy> so if distributions like Debian includes Prometheus, does it mean that they patch it themself after end of official support period?
2022-12-29 15:27:31 <pj> Debian is the type of distribution that will keep patching packages for 10 years straight
2022-12-29 15:28:31 <pj> seriously though, they definitely patch packages themselves if something is not supported anymore
2022-12-29 15:28:47 <raspbeguy> ok
2022-12-29 16:10:08 <pogduhog> Compiling custom kernel packages with ccache yields a low hit rate (about 2%). Is there a trick I am missing?
2022-12-29 16:12:41 <psykose> KBUILD_BUILD_TIMESTAMP has to be set  iirc
2022-12-29 16:15:59 <pogduhog> Looking at the APKBUILD I copied (linux-lts) it is already set: export KBUILD_BUILD_TIMESTAMP="$(date -Ru${SOURCE_DATE_EPOCH:+d @$SOURCE_DATE_EPOCH})"
2022-12-29 16:16:35 <pogduhog> I tried amending this to a static value as well but it had no affect
2022-12-29 16:21:47 <psykose> no idea then :)
2022-12-29 16:21:58 <psykose> i build my own all the time and that is the only thing set to static and ccache works for me
2022-12-29 16:22:13 <psykose> did you verify that ccache -s shows more misses? maybe it's not being used at all
2022-12-29 16:22:28 <psykose> it also gets invalidated if the compiler ever upgrades, etc
2022-12-29 16:22:41 <psykose> so you should reproduce this "now" with both builds at once
2022-12-29 16:24:52 <pogduhog> Yeah lots of misses and the cache grows so it is 'working'
2022-12-29 16:25:10 <pogduhog> What value do you use for that variable?
2022-12-29 16:27:32 <pogduhog> How do you mean both builds at once? 
2022-12-29 16:35:55 <psykose> i.e. build it
2022-12-29 16:36:00 <psykose> then just build it a second a time
2022-12-29 16:36:06 <psykose> a second time*
2022-12-29 16:36:10 <psykose> hmm
2022-12-29 16:36:14 <psykose> i just have it set to 1
2022-12-29 16:36:25 <psykose> ah no
2022-12-29 16:36:27 <psykose> empty string
2022-12-29 16:38:54 <pogduhog> Have been rebuilding a second time 
2022-12-29 16:39:20 <pogduhog> Will try setting it to an empty string myself, thanks
2022-12-29 18:37:35 <ddevault> pj: no particular reason; open a ticket to discuss
2022-12-29 18:49:03 <pj> discuss what? :)
2022-12-29 18:49:49 <pj> from what I see it, we should just pin it to next lts version once it becomes available
2022-12-29 18:50:02 <pj> words hard again
2022-12-29 18:50:11 <alpine-meetbot> pj meant to say: from what I see, we should just pin it to next lts version once it becomes available
2022-12-29 18:50:11 <pj> s/see it/see
2022-12-29 20:39:08 <raspbeguy> As much as I understand your opinion and think this is the wise thing to do, I'll be sad that I won't be able to use prometheus on edge anymore
2022-12-29 20:42:24 <psykose> what regression did you get hit by
2022-12-29 20:46:10 <raspbeguy> If we follow TLS only when it'll be released, then there won't be any regression
2022-12-29 20:47:38 <raspbeguy> I have no real argument for keeping latest version in alpine. I'll juste be sad to use a rolling release with some lts packages
2022-12-29 20:48:02 <raspbeguy> S/TLS/LTS
2022-12-29 20:48:02 <alpine-meetbot> raspbeguy meant to say: If we follow LTS only when it'll be released, then there won't be any regression
2022-12-29 20:49:59 <raspbeguy> Also it will make repology stats for alpine edge a bit less good :)
2022-12-29 20:52:12 <raspbeguy> But as I said I understand perfectly the reasons
2022-12-29 20:56:02 <raspbeguy> I believe that the next major version (or the one after it) will be LTS so we won't have to wait for long
2022-12-29 21:11:51 <psykose> if you never said anything nobody would even consider making it lts :p
2022-12-29 21:12:25 <psykose> personally i use victoriametrics so i have no horse in this one
2022-12-29 21:35:30 <psykose> the
2022-12-29 21:36:29 <ikke> So profound
2022-12-30 07:18:35 <pj> raspbeguy: someone still has to do the work, i just complain online :P
2022-12-30 07:35:50 <raspbeguy> psykose: well I originally wanted a different package for lts
2022-12-30 07:36:15 <raspbeguy> Still think this is the best option for edge users.
2022-12-30 08:37:50 <Hello71> pogduhog: you need to turn off gcc plugins
2022-12-30 09:07:19 <pogduhog> @Hello71 how do I go about doing that?
2022-12-30 09:07:49 <Hello71> CONFIG_GCC_PLUGINS=n
2022-12-30 09:07:51 <Hello71> or somesuch
2022-12-30 09:14:14 <pogduhog> I will give it a try now thanks 
2022-12-30 12:20:22 <PabloCorreaGomez[m]> ikke: thanks for the follow-up. I am basically offline until FOSDEM. So will not be able to make any use of it until then :)
2022-12-30 12:20:35 <ikke> np
2022-12-30 21:48:42 <ikke> psykose: deployed latest aports-qa-bot, please do let me know if it is misbehaving
2022-12-30 21:48:57 <psykose> ack, will keep an eye out
2022-12-30 21:49:10 <psykose> what changes to expect? i guess the group stuff is in
2022-12-30 21:49:57 <ikke> It still needs to be enabled
2022-12-30 21:50:04 <ikke> It uses the auth proxy now
2022-12-30 21:50:17 <psykose> okie
2022-12-30 21:50:22 <psykose> thanks
2022-12-30 21:50:55 <ikke> psykose: summary: https://gitlab.alpinelinux.org/alpine/infra/aports-qa-bot/-/blob/master/CHANGES
2022-12-30 21:52:53 <psykose> ay, warnprotectedbranch :)
2022-12-30 21:53:04 <psykose> if theres's also a warnprivate that would be nice
2022-12-30 21:53:13 <psykose> hm, what was the reason for strip +tag
2022-12-30 21:54:03 <ikke> Not entirely sure, but I suspect some users had user+tag@domain.org in the aports, while without it in gitlab
2022-12-30 21:54:18 <psykose> ah, for assign, right
2022-12-30 21:54:20 <psykose> yeah makes sense
2022-12-30 21:57:15 <ikke> Let me enable the 2 new services
2022-12-30 21:57:40 <ikke> There is also GreetFirstContributors and CommentOnApproval, do we want that?
2022-12-30 21:58:06 <psykose> would just be my personal opinion, hmm, probably no for the latter?
2022-12-30 21:58:16 <psykose> approval is already clear and generates an activity event
2022-12-30 21:58:20 <ikke> Yeah
2022-12-30 21:58:25 <psykose> for the former, idk
2022-12-30 21:58:42 <psykose> "no" if it's literally that, perhaps "yes" if it pointed to some documentation that would be useful for new people
2022-12-30 21:58:51 <psykose> but idk if we have something very good for the latter and it probably does not have that
2022-12-30 21:59:21 <ikke> https://gitlab.alpinelinux.org/alpine/infra/aports-qa-bot/-/blob/master/Services/GreetFirstContributors/definition.go#L15
2022-12-30 22:00:17 <psykose> that sure is friendly
2022-12-30 22:00:37 <psykose> i guess at least link to coding/commitstyle.md
2022-12-30 22:00:43 <psykose> but idk, up to you
2022-12-30 22:03:41 <ikke> I'll wait for now
2022-12-30 22:17:06 <ikke> PabloCorreaGomez[m]: fyi, it has been deployed
2022-12-31 07:41:37 <vkrishn> morning, re: aports-qa-bot is only for aports or irc help too ^ ?
2022-12-31 08:40:23 <knuxify> since when do we have a new alpine-qa-bot account, and why does it keep closing my mr claiming i made it from a protected branch
2022-12-31 08:40:33 <ikke> vkrishn: only gitlab
2022-12-31 08:40:46 <knuxify> (i checked and i have no protected branches)
2022-12-31 08:40:57 <ikke> knuxify: it used to be algitbot
2022-12-31 08:41:21 <ikke> knuxify: let me disable that paart
2022-12-31 08:41:47 <ikke> knuxify: sorry for that
2022-12-31 08:43:39 <knuxify> ikke: it's fine, no big deal :p
2022-12-31 08:48:41 <vkrishn> if anyone willing to do something similar for irc channels, then I can help build the text/statement mesh, https://tpaste.us/Oq78 
2022-12-31 09:12:14 <vkrishn> wait does "protected branch" means being pushed from devs "master" branch or a locked branch (that I don't understand), last I know merge request to master->master was dropped
2022-12-31 09:15:22 <vkrishn> ok got the answer
2022-12-31 09:15:51 <ikke> By default master is protected, so if they make a commit directly on master and make a merge request from that, the merge request cannot be rebased by the aports developers
2022-12-31 09:15:54 <vkrishn> https://gitlab.alpinelinux.org/alpine/infra/aports-qa-bot/-/blob/master/Services/WarnProtectedBranch/definition.go
2022-12-31 09:16:02 <vkrishn> yes, thanks
2022-12-31 09:17:14 <vkrishn> maybe it should be like "seems you have pushed merge request for either of these branches (master,a,b), which are currently protected, try some other branch"
2022-12-31 09:17:14 <ikke> Similar if the repo is private
2022-12-31 09:17:29 <vkrishn> merge request from^
2022-12-31 09:17:32 <ikke> And the MR should not be closed
2022-12-31 09:17:37 <ikke> That's just rude
2022-12-31 09:22:29 <vkrishn> yes, this is a good flow, branches are frugal/cheap on git unlike svn, and should be used whenever possible, I did a flow chart once, https://gist.github.com/insteps/7bd63123e195b40b23044596d4e4eced
2022-12-31 09:23:01 <vkrishn> think, I should update it to next version, with some new added practices I follow
2022-12-31 09:24:42 <ikke> It's a bit crowded
2022-12-31 09:27:00 <vkrishn> hmm.. ok, will make the graph objects smaller
2022-12-31 09:27:24 <vkrishn> and sparsely placed
2022-12-31 10:23:43 <psykose> nice to see i went to bed right when the bot planned its revenge
2022-12-31 10:23:44 <psykose> :D
2022-12-31 10:24:23 <ikke> :D
2022-12-31 11:01:48 <ikke> hmm, if I look at the WarnProtectedBranches service, I never see it checking if the branch the MR is from is actually protected? 🤨
2022-12-31 11:06:58 <ikke> psykose: why did you close !42767 aagain?
2022-12-31 11:07:11 <psykose> the branch is deleted
2022-12-31 11:07:13 <psykose> there's no changes
2022-12-31 11:07:21 <ikke> ah ok
2022-12-31 11:07:25 <psykose> a16bitsysop read the autoclose and deleted the branch after
2022-12-31 11:07:27 <psykose> much confusion
2022-12-31 11:07:31 <ikke> sigh
2022-12-31 11:07:38 <psykose> (the gitlab ui doesn't update for that until someone touches the mr)
2022-12-31 11:08:37 <a16bitsysop> it was created and pushed with glab, so thought maybe glab had done something wrong
2022-12-31 11:08:51 <psykose> nah, just the bot being silly
2022-12-31 11:09:10 <ikke> The entire service is just broken
2022-12-31 11:09:38 <ikke> It just adds that message and closes the MR for any MR
2022-12-31 11:09:59 <a16bitsysop> okay will open again
2022-12-31 11:10:09 <ikke> I now disabled it instead of setting it to DryRun
2022-12-31 11:13:12 <a16bitsysop> it just did it again when I cherry picked the commit from my master branch?
2022-12-31 11:15:24 <a16bitsysop> in my repository settings i have 0 protected branches
2022-12-31 11:16:15 <ikke> It should not do anything..
2022-12-31 11:17:29 <a16bitsysop> will create a fresh commit and try again
2022-12-31 11:19:08 <ikke> https://gitlab.alpinelinux.org/alpine/infra/aports-qa-bot/-/blob/master/Services/WarnProtectedBranch/definition.go does not seem to contain any logic to check protected branches
2022-12-31 11:22:24 <psykose> :D
2022-12-31 11:23:23 <ikke> lesigh
2022-12-31 11:47:07 <nmeum> can we add something along the lines of https://tpaste.us/Jro4 to tpaste.us? 🥺
2022-12-31 11:47:48 <nmeum> I haven't tested it but I strongly suspect that it should work :S
2022-12-31 11:48:37 <ikke> Should be possible
2022-12-31 11:48:40 <ikke> feel free to MR
2022-12-31 11:48:56 <psykose> that looks poggers
2022-12-31 11:53:17 <ikke> I have a rewrite in go added that as well
2022-12-31 12:01:45 <nmeum> ikke: https://gitlab.alpinelinux.org/nmeum/turbo-paste/-/merge_requests/1
2022-12-31 12:01:59 <nmeum> had to make some more adjustments to not break autodetection when &lang is not specified
2022-12-31 12:02:00 <ikke> You made it to your own fork :P
2022-12-31 12:03:49 <nmeum> m)
2022-12-31 12:03:50 <nmeum> right
2022-12-31 12:04:06 <nmeum> so used to just clicking the url in the git push motd
2022-12-31 12:05:18 <nmeum> fixed
2022-12-31 12:39:27 <vkrishn> go is preferred lang for al infra related codes ?
2022-12-31 12:39:44 <vkrishn> lua ?
2022-12-31 12:39:52 <ikke> It has my preference atm
2022-12-31 12:42:58 <ikke> ptrc: I removed the code for WarnProtectedBranch now
2022-12-31 12:44:51 <ptrc> thanks :)