2014-11-01 01:21:15 does alpine have any smart card tools? 2014-11-01 01:42:39 it does 2014-11-01 01:42:49 gpg includes support for smartcards 2014-11-01 01:43:00 i packaged libccid also, dunno if i published it 2014-11-01 01:57:50 grr 2014-11-01 01:57:58 it doesn't like this one security token I bought 2014-11-01 01:57:58 :P 2014-11-01 01:58:01 do you use a usb token? 2014-11-01 02:15:27 Mp5shooter: you mean a yubikey? 2014-11-01 02:17:59 I was looking at http://www.safenet-inc.com/multi-factor-authentication/authenticators/pki-usb-authentication/etoken-pro/ 2014-11-01 05:55:33 which package contains the certutil command? 2014-11-01 08:49:46 ncopa: hey 2014-11-01 15:02:55 I'm trying to setup Alpine as an explicit sslbump proxy and I can't get it to actually do ssl. I've followed all instructions on the wiki and can only get http connections to work. 2014-11-01 15:04:08 access.log doesn't show anything when I try say https://www.google.com from my browser that is set to use the proxy and the browser only shows internal communications error (using opera as a test browser). 2014-11-01 15:05:41 any hints on where to look for why its blocking would be greatly appreciated. This is a default install of alpine in a vm with just adding the packages from the wiki for squid and sslbump settings 2014-11-01 18:41:49 is it just me or is there no certutil or pk12util command in alpine's nss package? 2014-11-01 18:43:59 oh 2014-11-01 18:44:02 nss-tools has them 2014-11-03 08:28:22 <^7heo> moin @ 2014-11-03 08:28:35 <^7heo> what was the problem with Error relocating /usr/lib/firefox-33.0.2/libmozalloc.so: malloc_usable_size: symbol not found already? 2014-11-03 08:28:40 <^7heo> wrong versino of musl? 2014-11-03 08:50:43 <^7heo> damn, how to upgrade musl without pull the entire packageset? 2014-11-03 08:52:45 ^7heo, are you mixing edge and latest release? 2014-11-03 08:53:13 and yes, that sounds like firefox is from edge, but musl from 3.0-stable 2014-11-03 08:54:06 <^7heo> fabled: I don't know what you're answering to; but I just try to make firefox 33 working. 2014-11-03 08:54:15 <^7heo> and it can't be done without edge afaik 2014-11-03 08:55:21 <^7heo> So I'm going to upgrade to edge. 2014-11-03 08:56:04 <^7heo> fabled: and yes I was "mixing" latest and edge (and testing), using repo pinning 2014-11-03 08:56:18 <^7heo> but I got 4 packages that weren't from edge 2014-11-03 08:56:24 ok 2014-11-03 08:56:27 <^7heo> from latest I mean 2014-11-03 08:56:28 yes, that's the reason 2014-11-03 08:56:36 <^7heo> and 3 of them were from edge, one from testing 2014-11-03 08:56:47 <^7heo> all was working, the only thing that broke was ff 2014-11-03 08:56:54 <^7heo> because of the dep on musl-1.1.5 2014-11-03 08:56:55 would be nice if abuild tracked package versions for when symbols are introduced, and create automatically versioned dependencies 2014-11-03 08:57:12 but that would be quite a bit of work 2014-11-03 08:57:13 <^7heo> I'll see if edge has less problems with firefox 2014-11-03 08:57:18 works for me 2014-11-03 08:57:31 that is firefox on full edge install is working here 2014-11-03 08:57:31 <^7heo> I have some VERY memory intensive webpages 2014-11-03 08:57:35 <^7heo> that I have to work with 2014-11-03 08:57:53 <^7heo> rebooting 2014-11-03 10:05:44 <^7heo> did anyone here have problems with mplayer on edge 2014-11-03 10:05:45 <^7heo> ? 2014-11-03 10:05:49 <^7heo> it gets killed... 2014-11-03 10:06:45 <^7heo> right after set_tid_address 2014-11-03 11:24:38 ^7heo, you might want to try mpv 2014-11-03 11:25:08 <^7heo> I'm using mpg123 for now... 2014-11-03 11:25:23 <^7heo> but that's not fixing the mplayer error 2014-11-03 11:40:42 re: 3rd party keys, ubuntu has keyserver 2014-11-03 11:40:59 one does, sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys 2014-11-03 11:57:26 ^7heo: I use stable and have musl 1.1.5-r0 @edge pinned 2014-11-03 11:58:22 <^7heo> jomat: I wanted to do that too. 2014-11-03 11:58:35 <^7heo> I've tried "apk add musl@edge" and "apk add musl@testing" 2014-11-03 11:58:43 <^7heo> both were the same effect has "apk fix" 2014-11-03 11:58:57 <^7heo> (i.e. returning the number of installed packages and the disk space occupied) 2014-11-03 11:59:20 <^7heo> I didn't get anything to work using "apk add musl@ anything" 2014-11-03 11:59:27 <^7heo> how did you pin musl? 2014-11-03 11:59:34 ^7heo: That's a bug in apk: http://bugs.alpinelinux.org/issues/3493 2014-11-03 11:59:56 there are some dependencies which aren't displayed 2014-11-03 12:00:05 <^7heo> oh 2014-11-03 12:00:28 <^7heo> but 2014-11-03 12:00:41 I think it worked for me somehow by downloading the musl 1.1.5 package and installing it local or so 2014-11-03 12:00:55 <^7heo> "apk add musl=1.1.5-r0" did display a 100+ item list of upgrades to do. 2014-11-03 12:01:27 <^7heo> which didn't make sense to me, because what can work with 1.1.4-r3 can work with 1.1.5-r0... 2014-11-03 12:01:56 Uh 2014-11-03 12:04:19 <^7heo> what? 2014-11-03 12:06:22 <^7heo> http://sprunge.us/YDRN 2014-11-03 12:07:43 I don't remember if I also had to update so much, but I don't think so 2014-11-03 12:30:52 <^7heo> jomat: updating so much means pulling edge 2014-11-03 13:46:57 <^7heo> the URL in the info of the package 'vanessa_socket' is wrong. 2014-11-03 13:47:12 <^7heo> it says: http://horms.net/projects/perdition/ and should be: http://horms.net/projects/vanessa/ 2014-11-03 14:14:52 <^7heo> the package dante has to be patched, also, but upstream, I believe. 2014-11-03 14:20:48 afternoon 2014-11-03 14:21:35 <^7heo> ScrumpyJack: moin. 2014-11-03 19:32:00 argh, great, plugging in a PS3 controller freezes the kernel a few seconds later /o\ 2014-11-03 19:37:15 :( 2014-11-03 19:58:15 <^7heo> jomat: why so much hate? 2014-11-03 19:58:28 <^7heo> jomat: can't you get a microsoft xbox controller like everybody else? :D 2014-11-04 08:33:24 morning 2014-11-04 08:37:37 morning 2014-11-04 08:44:27 morning 2014-11-04 10:17:07 morning 2014-11-04 11:22:26 upgrading from 3.0_beta2 to last stable is posible? should this be enough "apk update && apk upgrade" 2014-11-04 13:53:10 <^7heo> crow: moi 2014-11-04 13:53:11 <^7heo> n 2014-11-04 14:03:29 hi :) 2014-11-04 14:10:46 <^7heo> oh I meant 2014-11-04 14:10:49 <^7heo> clandmeter: moin 2014-11-04 14:10:51 <^7heo> but hi crow 2014-11-04 14:14:15 does 'apk upgrade -U' tell you when it upgrades things or do you need '-v' ? 2014-11-04 14:15:34 ^7heo: good afternoon 2014-11-04 14:16:21 <^7heo> thx 2014-11-04 14:18:05 buckley310 i used this "apk update && apk upgrade --update-cache --available" this was result http://dpaste.com/356X8S3 , system reboot was just fine 2014-11-04 14:21:28 hmmm --available isnt in the "--help" output, whats that? 2014-11-04 14:27:08 buckley310 it is in "apk --help upgrade" 2014-11-04 14:27:27 almost on the end. 2014-11-04 14:28:07 i see. 2014-11-04 14:28:17 okay thanks, ill try that next time. 2014-11-04 14:28:24 each applet has its own little help 2014-11-04 15:02:05 <^7heo> the use of "applet" in that context really surprised me. 2014-11-04 15:02:14 <^7heo> but it turns out to be a valid use. 2014-11-04 17:42:39 Hi ,is there a hidden wiki corner for a kind of fluxbox-flavour of alpine-linux (with lxpanel conky-all deluge qemu-launcher wicd qupzilla)?cant find these in the package browser ,im not optimistic enough for compiling those i guess.zulucrypt would be nice too as luks-frontend ... thanks 2014-11-04 17:43:07 hey making apkbuilds is easy, just go ahead and do it! 2014-11-04 17:51:11 hmmmm , last time it didnt work for me ... maybe i try again .the (only)other thing that stopped my trial was the autoconfiguration of sound /restart alsa,when restarting on a different machine. would it be complicated to import the soundconfig scripts of e.g. grml (www.grml.org)? works always for me ,thanks 2014-11-05 04:37:52 hey all 2014-11-05 07:13:11 hi 2014-11-05 07:16:20 moin 2014-11-05 08:44:51 <^7heo> moin ncopa 2014-11-05 08:45:37 hi 2014-11-05 08:46:26 <^7heo> ncopa: is there a simple way to solve the problem of the missing cherry hid code in the initramfs? 2014-11-05 08:47:07 ^7heo: i think the only way is to include the module in mkinitfs's features.d modules file 2014-11-05 08:54:59 morning 2014-11-05 09:00:26 <^7heo> ncopa: okay,thanks for confirming. 2014-11-05 09:09:34 ^7heo: http://git.alpinelinux.org/cgit/mkinitfs/commit/?id=bd763d9334f703cdef87fe3c5973c71f5eafc101 2014-11-05 09:10:43 <^7heo> ncopa: many thanks :) 2014-11-05 09:11:08 <^7heo> so when I update my alpine, in some time, I'll have the USB support ;) 2014-11-05 09:11:09 <^7heo> \o/ 2014-11-05 09:11:15 yup 2014-11-05 09:11:20 with v3.1 2014-11-05 09:11:23 <^7heo> or edge. 2014-11-05 09:11:26 <^7heo> but yeah. 2014-11-05 09:11:41 <^7heo> many many thanks. 2014-11-05 09:11:56 <^7heo> I'm basically using PS/2 anyway 2014-11-05 09:24:05 <^7heo> for obvious security reasons. 2014-11-05 09:24:12 <^7heo> (and performance too) 2014-11-05 14:29:25 Tervehdys! Hilsener! Grüße! Greetings! 2014-11-05 15:00:26 <^7heo> moin 2014-11-05 16:43:42 I was considering installing Alpine on a desktop pc. I noticed that it lacks some packages I like to use. E.g., clementine, gvim. 2014-11-05 16:44:20 Is there way to easily package these? Something like Arch's PKGBUILDs? 2014-11-05 16:59:44 yup 2014-11-05 17:00:12 http://wiki.alpinelinux.org/wiki/Creating_an_Alpine_package 2014-11-05 19:30:31 hello is what is current status of pie on alpine? it's mostly enabled or disabled in packages? 2014-11-05 19:31:24 hello algitbot robot 2014-11-05 19:37:07 i think PIE is default and used in most packages 2014-11-05 19:37:30 readelf -h will show you if a package is PIE 2014-11-05 19:37:31 look foe 2014-11-05 19:37:36 Type: DYN (Shared object file) 2014-11-05 19:37:44 dalias: amazing 2014-11-05 19:37:59 do you know if anybody is using alpine as desktop? 2014-11-05 19:38:51 at least some of the developers do.. otherwise there would be no point of the firefox package 2014-11-05 19:39:25 nsz: amazing 2014-11-05 19:39:45 one last thing i am curious about i see lots of packages on packages pages 2014-11-05 19:40:07 does it mean all those packages use musl for libc instead glibc? 2014-11-05 19:40:39 alpine is musl only 2014-11-05 19:40:50 incredible work guys 2014-11-05 19:41:23 i dont think it makes sense to have dual libc setup because then you would need all lib dependencies to have twice for the two libcs 2014-11-05 19:42:12 well if you don't care about size and memory overhead and just want lots of stuff to work, it can make sense. see: bedrock linux that's going to have one libc (and whole set of libs) per distro you're mixing :) 2014-11-05 19:42:27 it makes more sense to just resort to a chroot with an existing glibc distro than to do a dual libc setup 2014-11-05 19:42:42 nsz: no i mean i thought things depend on glibc so hard it cannot be removed for certain projects - but i for example see qt <- so it means in the end c parts are musl 2014-11-05 19:42:46 right? 2014-11-05 19:43:54 okay last thing do you sign your packages? 2014-11-05 19:43:54 qt certainly doesn't depend on glibc 2014-11-05 19:43:59 it runs on windows even 2014-11-05 19:44:14 dalias: roger 2014-11-05 19:44:22 eg qt depends on libxml2 (because everything depends on libxml2 for some reason) if alpine liked qt against glibc then it would need to package libxml2-glibc too 2014-11-05 19:44:51 and same for all the dependencies 2014-11-05 19:45:00 wow there is even gtk 2014-11-05 19:45:24 so one could build really normal desktop 2014-11-05 19:46:05 okay if it is so why more distros are not using musl isn't it much better designed than glibc? 2014-11-05 19:47:09 because it's all or nothing, everything depends on libc and if there is one big app that's glibc only then you need all the libs twice 2014-11-05 19:47:32 nsz: i see 2014-11-05 19:47:36 and musl is new 2014-11-05 19:47:46 thank you guys this is awesome actually 2014-11-05 21:34:42 <^7heo> I use alpine as a desktop 2014-11-05 21:35:32 <^7heo> and about the dual libc, if alpine uses glibc once, I'm not going to touch it ever again. 2014-11-05 21:35:56 haha 2014-11-05 21:36:09 <^7heo> if I need the glibc, there's docker/lxc for that 2014-11-05 21:36:54 :) 2014-11-05 21:37:02 <^7heo> but yeah, grsec + PAX + musl is a very great combo. 2014-11-05 21:37:26 <^7heo> it's not working everywhere, because of the GNUisms in all the code around us... 2014-11-05 21:37:39 <^7heo> but with time we will be able to patch it I believe. Upstream I mean. 2014-11-05 21:37:40 and the whole pax-being-broken thing :) 2014-11-05 21:37:48 <^7heo> it's not SO broken 2014-11-05 21:39:20 dalias: I will have to say that PaX and grsecuritys revision control is less to be than desirable 2014-11-05 21:40:06 <^7heo> systmkor: what do you mean? 2014-11-05 21:40:20 ^7heo: it's basically they manually throw it in a folder 2014-11-05 21:40:23 annnd that's about it 2014-11-05 21:41:02 I am emailing the PaX team at the moment to figure out what they are doing manually that is not reproducable in a revision control system 2014-11-05 21:41:49 :) 2014-11-05 21:42:16 I do think they have a unique challenge compaired to common software engineering rcs process 2014-11-05 21:42:41 hey i am building a multiboot USB with grub 2 so I can boot iso's 2014-11-05 21:42:57 I can't seem to get the booting process of Alpine quite right 2014-11-05 21:43:29 I am also going trying to be able to boot openbsd56 2014-11-05 21:43:37 any thoughts or where I should look to figure it out 2014-11-05 21:43:48 systmkor, what is their unique challenge? 2014-11-05 21:44:14 well I dropped my other laptop that has a working CD burner 2014-11-05 21:44:20 and I have one USB at the moment :P 2014-11-05 21:44:32 soo basically multiboot install disks 2014-11-05 21:51:53 <^7heo> right now I'm trying to start Steam on Alpine 2014-11-05 21:52:00 <^7heo> as a docker container. 2014-11-05 21:52:11 <^7heo> more info on that if I make it work 2014-11-05 21:52:14 interesting 2014-11-05 21:52:30 dalias: any suggestions? 2014-11-05 21:53:14 i'm not particularly good with that kinda stuff 2014-11-05 21:53:30 still trying to figure out how to boot my dragonboard and install alpine/arm on it 2014-11-05 21:53:47 <^7heo> systmkor: what's the problem? 2014-11-05 21:54:08 i either get a sys init error or a kernel panic 2014-11-05 21:54:14 lemme post my grub.cfg one sec 2014-11-05 21:54:18 <^7heo> why grub? 2014-11-05 21:54:30 <^7heo> is there a valid use case, or is it just masochism? 2014-11-05 21:54:39 https://gist.github.com/7adf17724411dbd79dab 2014-11-05 21:54:44 booting from iso files 2014-11-05 21:54:50 unless syslinux can do that? 2014-11-05 21:55:03 i am also somewhat familiar with u-boot 2014-11-05 21:55:47 <^7heo> nah I think that syslinux is missing the ISO fs 2014-11-05 21:56:01 <^7heo> but at least you're using grub1. 2014-11-05 21:56:12 <^7heo> or not... 2014-11-05 21:56:32 <^7heo> that's grub2 simplified to the bare minimum, isn't it? 2014-11-05 21:56:41 well sorta 2014-11-05 21:56:52 grub2 sorta has way too many features sorta 2014-11-05 21:56:56 i am using grub2 2014-11-05 21:57:06 you can still use grub1 style configuration though 2014-11-05 21:57:07 <^7heo> yeah 2014-11-05 21:57:24 <^7heo> yeah, with "style" being subject to personal interpretation 2014-11-05 21:57:26 <^7heo> but yeah okay 2014-11-05 21:57:38 <^7heo> I could help much more with grub 1 I think 2014-11-05 21:58:01 well what I think I have posted should be somewhat similar to grub 1 config 2014-11-05 21:59:24 <^7heo> systmkor: have you seen isolinux? 2014-11-05 21:59:41 i know of it but not that familiar with it 2014-11-05 22:00:21 however I thought isolinux was for creating bootable isos 2014-11-05 22:00:24 not booting isos 2014-11-05 22:00:31 <^7heo> hmm 2014-11-05 22:00:38 <^7heo> I didn't read enough to be sure 2014-11-05 22:00:45 by that I mean like a bootloader that then "chainloads" into the iso file 2014-11-05 22:02:53 really the goal is that I can boot an iso without having to do any of the boolshit of figuring out its initrd and kernel image 2014-11-05 22:03:02 and ignore all the variations of booting linux 2014-11-05 22:03:15 <^7heo> well, I don't know enough about bootloaders and isofs to help. 2014-11-05 22:03:27 <^7heo> and I really want to try to get a steam working on my alpine 2014-11-05 22:03:56 ^7heo: I would actually be down to help 2014-11-05 22:04:03 I am wanting to have a good guide on the wiki 2014-11-05 22:04:10 for creating a lxc 2014-11-05 22:04:12 <^7heo> for steam? 2014-11-05 22:04:14 <^7heo> ah 2014-11-05 22:04:18 <^7heo> docker is so freaking easy to dl 2014-11-05 22:04:22 but my goals is just lxc as a sandbox of an app 2014-11-05 22:04:23 systmkor: If your iso can fit into memory, you can use syslinux's memdisk. 2014-11-05 22:04:25 <^7heo> and use 2014-11-05 22:04:26 <^7heo> it's like 2014-11-05 22:04:29 <^7heo> plug and play 2014-11-05 22:04:39 not so much a lightweight distro emulation 2014-11-05 22:05:11 <^7heo> docker? 2014-11-05 22:07:29 well more of how it is used 2014-11-05 22:07:37 so commonly people will emulate a full stack in docker 2014-11-05 22:08:04 basically goal for me is to use lxc more strictly as a sandbox and networking controll 2014-11-05 22:08:17 so for example in alpine 2014-11-05 22:08:25 create a chroot directory 2014-11-05 22:08:31 install the app you want in that directory 2014-11-05 22:08:38 then get lxc to boot that directory 2014-11-05 22:08:41 and run only that app 2014-11-05 22:09:01 without busybox or anything that isn't strictly needed for that application or service 2014-11-05 22:12:07 well maybe supervisord or something to controll or maintain process 2014-11-05 22:12:13 idk entirely yet 2014-11-05 22:18:17 ^7heo: so where are you at with getting steam working? 2014-11-05 22:58:07 <^7heo> systmkor: I got hooked to dr who 2014-11-05 22:58:12 lol 2014-11-05 22:58:17 <^7heo> and now it's too late to work 2014-11-05 22:58:49 <^7heo> plus, I'm rather far away from the wifi now 2014-11-05 22:58:54 <^7heo> and it's vERY weak 2014-11-05 23:01:54 you must construct additional access points 2014-11-05 23:07:57 <^7heo> nah 2014-11-05 23:08:01 <^7heo> I must plug my laptop 2014-11-05 23:08:13 <^7heo> I'm @home 2014-11-05 23:08:18 <^7heo> not on the move. 2014-11-05 23:13:17 ^7heo: still awake? 2014-11-05 23:20:17 <^7heo> clandmeter: yeah 2014-11-05 23:20:19 <^7heo> clandmeter: why? 2014-11-05 23:20:41 <^7heo> I was watching some video of a guy being overly excited about his own semence. 2014-11-05 23:20:42 i dont know. 2014-11-05 23:20:54 <^7heo> (Hy, if you wonder) 2014-11-05 23:21:18 <^7heo> like, the guys implements lisp in python, and is amazed at "how much complicated code it generates" 2014-11-05 23:21:20 i thought you went to bed earlier most of the time. 2014-11-05 23:21:49 <^7heo> I'm depressed, under pressure, lonely and stressed by more than one problem at once. 2014-11-05 23:21:54 <^7heo> So, yes, I'm in bed. 2014-11-05 23:22:12 <^7heo> Now, will I sleep? 2014-11-05 23:22:35 yes. and go for a swim tomorrow so you can spend 2-3 minutes without all the problems. 2014-11-05 23:22:58 <^7heo> or 2014-11-05 23:23:03 <^7heo> I can go for a swim 2014-11-05 23:23:11 <^7heo> and solve all my problems at once. 2014-11-05 23:23:46 swimming is when you're actively aflot. 2014-11-05 23:23:47 +a 2014-11-05 23:24:18 <^7heo> meh, that's semantic details. 2014-11-05 23:24:43 thats called drowning, not swimming. 2014-11-05 23:25:19 <^7heo> close enough 2014-11-05 23:25:29 <^7heo> and you're never seen me swimming. 2014-11-05 23:25:39 <^7heo> hem, sorry, according to your definition, drowning. 2014-11-05 23:28:48 <^7heo> okay, so, docker isn't accepting flags anywhere in the command line. 2014-11-05 23:28:50 <^7heo> which is good. 2014-11-05 23:31:46 <^7heo> anyway 2014-11-05 23:31:59 <^7heo> I'm going to have to do my own steam on docker... 2014-11-05 23:32:14 <^7heo> I just fear that steamguard is going to annoy me a LOT 2014-11-05 23:32:28 <^7heo> but 2014-11-05 23:32:32 <^7heo> if I'm lucky 2014-11-05 23:32:43 <^7heo> I can play my games on my laptop, from my desktop 2014-11-05 23:32:45 <^7heo> via the network 2014-11-05 23:32:51 <^7heo> on Alpine, running docker, running steam. 2014-11-05 23:33:14 <^7heo> Or I could just dual boot windows, since my laptop has a GTX 860 M 2014-11-05 23:34:29 <^7heo> anyway 2014-11-05 23:34:34 <^7heo> time to look at the ceiling 2014-11-05 23:34:36 <^7heo> o 2014-11-05 23:34:38 <^7heo> o/ 2014-11-05 23:36:45 gnite 2014-11-05 23:37:03 <^7heo> danke 2014-11-05 23:40:39 graag gedaan 2014-11-06 00:05:34 has anyone done full disk encryption with dm-crypt before? 2014-11-06 00:06:09 if I have a boot partition seperate from the root partition 2014-11-06 00:06:27 I would decrypt the root partition during the sysv / openrc init 2014-11-06 01:55:48 do you guys have a preference between chrony or openntpd? 2014-11-06 01:56:33 i prefer busybox ntpd 2014-11-06 01:57:45 it's less aggressive with regard to waking up and probing the time too often 2014-11-06 01:58:08 but best of all it treats leap seconds as clock drift rather than discontinuously fucking with the clock 2014-11-06 03:31:31 ncopa: oprofile requires some constants not in ftw.h (i.e. FTW_STOP, FTW_CONTINUE, etc.) 2014-11-06 03:31:39 also its arch is set to "" 2014-11-06 03:31:48 maybe b/c it doesnt work 2014-11-06 03:46:31 ncopa: ntop also has build errors and has its arch set to "" 2014-11-06 03:47:08 im not sure ntop or oprofile ever built 2014-11-06 03:55:48 at least not since the move to musl 2014-11-06 03:55:56 ufsutils also has no arch set 2014-11-06 03:58:07 i'm not sure what those FTW_* do 2014-11-06 04:01:48 oops i meant to put this in devel 2014-11-06 04:01:49 oh well 2014-11-06 04:01:52 lol 2014-11-06 04:01:57 no prob 2014-11-06 04:02:04 yeah, i dont either 2014-11-06 04:02:17 i think some of the packages are holdovers from the uclibc days 2014-11-06 04:02:27 uclibc has those FTW_ constants, but musl does not 2014-11-06 04:03:29 well if they represent nonstandard features in the ftw functionality, just adding the constants wouldnt help 2014-11-06 04:04:04 yeah, im not messing w/ it 2014-11-06 04:04:28 im just trying to do the easy stuff so others can skip it 2014-11-06 04:04:33 low hanging fruit 2014-11-06 06:58:22 <^7heo> ncopa: are you also going to say that I should use autotools to build that: https://github.com/bramp/archivemount ? 2014-11-06 06:58:30 <^7heo> ('cause, yeah, it's what they use) 2014-11-06 07:05:05 morning 2014-11-06 07:05:48 <^7heo> moin 2014-11-06 07:06:29 <^7heo> why are people using GNUisms THAT much? 2014-11-06 07:06:37 <^7heo> It's everything but portable... 2014-11-06 07:07:21 <^7heo> and I just found c99 features that aren't supported by gcc -std=c99 =/ 2014-11-06 07:08:56 <^7heo> and now I also have system includes having conflicting declarations... 2014-11-06 07:09:21 <^7heo> so much shit for only one file of c... 2014-11-06 07:09:52 <^7heo> there are days, I'd rather die than read this. And there are days everyday. 2014-11-06 07:10:29 ^7heo: yes use autotools if upstream uses it 2014-11-06 07:10:58 <^7heo> except that its only one file of c, and autotools is just there to generate 3 flags or something. 2014-11-06 07:11:17 yes, use it 2014-11-06 07:11:24 i agree its stupid 2014-11-06 07:11:35 <^7heo> :'( 2014-11-06 07:12:08 but rewriting build scripts for every planet on the earth is not worth it 2014-11-06 07:12:32 every project* (i shouldnt talk inphone a write on irc at same time) 2014-11-06 07:12:58 <^7heo> okay, maybe I'm missing something here 2014-11-06 07:13:06 <^7heo> in the git repo 2014-11-06 07:13:10 <^7heo> I have to 2014-11-06 07:13:21 <^7heo> aclocal && autconf && ./configure && make, right? 2014-11-06 07:13:53 <^7heo> 'cause configure is complaning about not finding install.sh or shtool or stuff 2014-11-06 07:16:45 <^7heo> ncopa: do you agree with the use of those commands: http://askubuntu.com/a/27679 ? 2014-11-06 07:16:55 <^7heo> with the --force flags and shit 2014-11-06 07:17:09 <^7heo> (I left out the first one, it worked) 2014-11-06 07:22:41 looks ok 2014-11-06 07:23:24 <^7heo> so much for "just works" 2014-11-06 07:23:28 libtoolize --force && aclocal -I m4 && autoheader && autoconf && automake --add-missing 2014-11-06 07:23:32 <^7heo> but ok I'll do the apk with that. 2014-11-06 07:23:33 normally works 2014-11-06 07:23:35 thanks 2014-11-06 07:23:40 we have lots of those 2014-11-06 07:24:07 <^7heo> yeah I can imagine 2014-11-06 07:24:24 <^7heo> I mean, most software is shit, so... 2014-11-06 07:25:05 even if its dog slow, and huge, it is kind of a standard 2014-11-06 07:25:18 <^7heo> I know 2014-11-06 07:25:19 which provides some level of consistency 2014-11-06 07:25:20 <^7heo> like 2014-11-06 07:25:30 <^7heo> C++, Java and Ruby are the standard now. 2014-11-06 07:25:32 which is nice when you have to deal with 1000+ packages 2014-11-06 07:25:36 <^7heo> Ubuntu and systemd are the standard now 2014-11-06 07:25:44 yeah :-/ 2014-11-06 07:26:02 <^7heo> can I make "suicide" an accepted standard too? 2014-11-06 07:26:11 no. you can not 2014-11-06 07:26:44 worst with autotools is when upstream developer have no clue about it 2014-11-06 07:26:49 and creates a mess of it 2014-11-06 07:27:09 and mix in gnulib 2014-11-06 07:27:39 if you find a messy autoconf script+buggy gnulib 2014-11-06 07:27:44 thats nightmare 2014-11-06 07:28:09 <^7heo> isn't that standard? 2014-11-06 07:28:33 no 2014-11-06 07:28:36 <^7heo> I mean, nightmares, mess, and shit? 2014-11-06 07:28:55 there are different levels of mess 2014-11-06 07:29:01 <^7heo> s/mess/hell/ 2014-11-06 07:29:07 most people are able to figure it out decently 2014-11-06 07:30:13 <^7heo> maybe yeah 2014-11-06 07:30:32 ^7heo: how goes the lxc challenge? 2014-11-06 07:30:46 <^7heo> systmkor: I needed archivemount in the middle 2014-11-06 07:30:53 ? 2014-11-06 07:31:09 <^7heo> ncopa: I'm just overly sad to see what something that could have been incredibely neat and well designed has turned in such a nightmare... 2014-11-06 07:31:17 <^7heo> "Just because we can" ™ 2014-11-06 07:31:30 <^7heo> incredibly even 2014-11-06 07:33:02 do you call https://github.com/bramp/archivemount/blob/master/configure.ac a nightmare? :) 2014-11-06 07:33:06 its not even close... 2014-11-06 07:33:45 <^7heo> Nah I'm not calling that a nightmare 2014-11-06 07:33:45 but hum 2014-11-06 07:33:54 no release tags? 2014-11-06 07:34:09 <^7heo> I didn't look 2014-11-06 07:34:18 <^7heo> oh 2014-11-06 07:34:20 <^7heo> you're right... 2014-11-06 07:34:26 http://www.cybernoia.de/software/archivemount/ 2014-11-06 07:35:27 ncopa: I was wondering what the decision process is for the kernelconfig of the linux-grsec kernel 2014-11-06 07:36:30 I just did a fresh install of 3.0.6 and was see what lxc-checkconfig would spit out with the default kernel and almost everything was missing or disabled 2014-11-06 07:36:37 i am not saying it has to be changed 2014-11-06 07:36:45 just curious about what makes the cut off 2014-11-06 07:37:17 nowdays its mostly like, disable by default and enable when someone asks 2014-11-06 07:37:40 i think i did look into the lxc config options some time ago 2014-11-06 07:38:09 and some of them had a price in memory consumption or performance 2014-11-06 07:38:23 i think i tried diable the most expensive features 2014-11-06 07:38:30 m'kay 2014-11-06 07:38:44 tried to find the balance cost/benefit 2014-11-06 07:38:48 i was wondering if it would make sense to have potentially linux-pax, and like linux-grsec 2014-11-06 07:39:02 were linux-grsec became the more full blown kernel security features enabled 2014-11-06 07:39:13 and linux-pax was the leaner minimal pax flags and such 2014-11-06 07:39:18 hm 2014-11-06 07:39:33 <^7heo> ncopa: is the link from the git repo? 2014-11-06 07:39:44 i have been thinking of only do linux-pax 2014-11-06 07:39:58 i would prefer keep number of different kernels down 2014-11-06 07:39:59 I will be going through a decent amount of the kernel flags for lxc, cgroups, pax, and grsec 2014-11-06 07:40:05 so I can do a write up 2014-11-06 07:40:09 as kernels are much maintenance work 2014-11-06 07:40:09 uefi 2014-11-06 07:40:12 :3 2014-11-06 07:40:36 systmkor: i have also been thinking of a linux-xen for dom0 2014-11-06 07:41:04 but i doubt i have the capacity to maintain all different variants 2014-11-06 07:41:11 I wish I could test that but the only spare servers I have don't have hardware virtualization 2014-11-06 07:41:15 ^7heo: yes, i found it on the git repo 2014-11-06 07:41:31 ^7heo: https://github.com/bramp/archivemount 2014-11-06 07:41:39 ncopa: well if nothing else a nice wiki with sections for xen, pax, grsec, lxc, cgroups 2014-11-06 07:41:42 would be nice 2014-11-06 07:41:45 at the top it says: 2014-11-06 07:41:45 FUSE filesystem using libarchive 2014-11-06 07:41:45 http://www.cybernoia.de/software/archivemount/ 2014-11-06 07:41:50 I'll see what I can whip up for that 2014-11-06 07:42:03 systmkor: would be great 2014-11-06 07:42:33 I sent in the mailing list, for thoughts about the yocto project but I didn't seem to get any feedback for that 2014-11-06 07:42:45 systmkor: i have also been thinking of adding different kernel configs for grsec patches kernel 2014-11-06 07:42:59 eg different builds from same patched kernel source 2014-11-06 07:43:28 its just that its very timeconsuming to maintain different kernel configs 2014-11-06 07:43:56 <^7heo> btw 2014-11-06 07:44:03 <^7heo> I have permission rights with fuse 2014-11-06 07:44:06 <^7heo> am I missing a group? 2014-11-06 07:44:12 <^7heo> I'm in fuse already I believe. 2014-11-06 07:44:15 ncopa: i could see that 2014-11-06 07:45:01 ^7heo: have a look at this: http://sprunge.us/gBJT 2014-11-06 07:45:50 other than yocto, nix package manager looks interesting 2014-11-06 07:45:56 <^7heo> ncopa: you're faster than I am 2014-11-06 07:47:34 ^7heo: I cheat. ok if i push it? 2014-11-06 07:48:18 <^7heo> ncopa: ofc 2014-11-06 07:48:30 <^7heo> it has to be done, not necessarily by me. 2014-11-06 07:48:45 <^7heo> btw, what about this fuse permission problem, any clue? 2014-11-06 07:55:02 <^7heo> mount("archivemount", "/path/to/file", "fuse.archivemount", MS_NOSUID|MS_NODEV, "fd=6,rootmode=40000,user_id=1000"...) = -1 EPERM (Operation not permitted) 2014-11-06 07:55:06 <^7heo> =/ 2014-11-06 07:55:13 <^7heo> and I'm in the group fuse 2014-11-06 08:12:37 Good Morning everyone 2014-11-06 08:12:55 morning …. 2014-11-06 08:15:28 ginjachris: goodnight :P 2014-11-06 08:15:35 or goodevening 2014-11-06 08:16:38 systmkor: :D LOL … youre right there… 2014-11-06 08:16:55 api984: ? 2014-11-06 08:17:47 what do you mean? 2014-11-06 08:17:59 systmkor: morning or night at your place? 2014-11-06 08:18:09 night 2014-11-06 08:18:14 morning here 2014-11-06 08:18:25 right 2014-11-06 08:18:47 :D 2014-11-06 08:18:58 systmkor: using alpine linux long? 2014-11-06 08:19:13 not that long maybe on an off for a month 2014-11-06 08:19:46 i like it 2014-11-06 08:19:56 systmkor: thats cool. did you implement it somewhere ? 2014-11-06 08:20:04 i have it running on a linode instance 2014-11-06 08:20:12 and a server in the office's utility closet 2014-11-06 08:20:22 *our office's 2014-11-06 08:20:24 systmkor: just to say. this is the BEST small APPLIANCE distro I ever used 2014-11-06 08:20:40 cool 2014-11-06 08:20:53 yah I think it's what the base of distros should be 2014-11-06 08:20:54 i setup a dovecot proxy, 2 smtp server with haproxy in HA mode 2014-11-06 08:21:12 some sort of a smtp failover 2014-11-06 08:21:16 nice 2014-11-06 08:21:35 yah I think since Alpine seems to be aimed at lean, secure, and reliable 2014-11-06 08:21:36 nice thing is ARP ip failover 2014-11-06 08:21:41 and the community is small 2014-11-06 08:21:44 yes 2014-11-06 08:21:53 ther could be some pretty intersting tecnical accomplishments made 2014-11-06 08:21:57 however no matter the version of alpine i can make it work 2014-11-06 08:22:00 *could be made 2014-11-06 08:23:25 right now I am trying to figure out how to make a generic way 2014-11-06 08:24:04 of installing a service like a git server of installing inside a lxc container 2014-11-06 08:24:10 with minimal manual configuration 2014-11-06 08:24:53 also not using lxc as a para-kvm virtualization which is neet 2014-11-06 08:25:11 it would also be nice to make it easier to work as a jail for a program 2014-11-06 08:27:10 api984: do you just use it a an stmp server? 2014-11-06 08:29:39 sry 2014-11-06 08:29:46 was on linux channel now 2014-11-06 08:30:20 systmkor: stmp? 2014-11-06 08:30:46 systmkor: i did not try LXC, sandboxing or docker or similar…. 2014-11-06 08:30:47 *smtp 2014-11-06 08:31:01 systmkor: smtp server yes 2014-11-06 08:31:27 systmkor: i did not do any hardcore configuration except ip failover 2014-11-06 08:32:10 does anybody have any experience with intel atom n2600 + gma 3600 graphics? i have a giada mini-pc with those and i can not get it to boot alpine linux 3.0.6. i get the "blank screen after display driver is loaded" which is mentioned in many google hits, but none of the mentioned fixes seem to work. afaik there should be a good (open) driver for it since linux 3.4 something. 2014-11-06 08:32:27 no sorry 2014-11-06 08:32:42 where can i easily see what is enabled in the grsec or vanilla kernels? 2014-11-06 08:32:50 hute: nop… any chance to ssh in if blank screen 2014-11-06 08:33:17 i didnt get either of those to work. ubuntu ofcourse boots without problems, but theyve had some proprietary support for the chip since 12.04 so it probably uses that driver. 2014-11-06 08:33:43 api984: i dont know. does the install iso enable sshd+dhcp by default? 2014-11-06 08:33:43 hute: uhm its in the aports under /main/linux-grsec/kernelx.conf 2014-11-06 08:33:59 ive just tried with the install iso so far 2014-11-06 08:34:07 hute: http://git.alpinelinux.org/cgit/amanison/aports/ 2014-11-06 08:34:22 systmkor: thanks 2014-11-06 08:34:30 hute: what are you booting off of? 2014-11-06 08:35:27 i have a 8GB usb stick where i installed the iso using unetbootin, so its syslinux i think 2014-11-06 08:36:58 hute: not to say that you shouldn't ue unetbootin but I have never personally have been able to get it to work 2014-11-06 08:37:10 *been able to get it to work 2014-11-06 08:37:52 hute: its been a while since i last configured it 2014-11-06 08:37:59 hute: not sure, dammm 2014-11-06 08:38:03 it does boot fine until i get the a line about "loading device drivers" somewhere around after udev is initialized. 2014-11-06 08:39:06 systmkor: i used unetbootin, because it was mentioned as the first alternative in the alpine linux wiki http://wiki.alpinelinux.org/wiki/Create_a_Bootable_USB 2014-11-06 08:39:11 hute: you can try a hack then… install as a VM… block these drivers in a blacklist if you know what their names are… or strip them out… image the VM and restore on barebone 2014-11-06 08:39:32 hute: yah I want to get rid of that, sorry rather new to this community myself 2014-11-06 08:40:02 systmkor: similar page in arch linux wiki recommends straight dd 2014-11-06 08:40:11 yah one sec 2014-11-06 08:40:32 on the unetbootin page 2014-11-06 08:40:34 check under manual 2014-11-06 08:41:13 api984: im hoping for something like that, but i dont know exactly what to blacklist. i guess ssh into the blank screen box and the blacklisting in VM or directly at the boot promt might be my best options 2014-11-06 08:42:43 i would like to use the official releases with as few modifying steps as possible, so maybe directly on the boot prompt 2014-11-06 08:44:29 hute: agreed… does it blackout on any other distro… 2014-11-06 08:46:33 api984: arch blacked out at exactly the same point as well. ubuntu (server 14.10) is the only monster ive got to boot up, but i really really do not want to install ubuntu on a light "hazzle" free appliance in the middle of debian systemd initd wars. im sure it will break nicely when the systemd change comes. 2014-11-06 08:47:33 i tried all the boot options like nomodeset i915.modeset=0 i915.blacklist=yes gma500_gfx.blacklist=yes etc i could google up 2014-11-06 08:50:41 ok if ubuntu boots up… use lsmod 2014-11-06 08:50:51 find out what modules are there regarding VGA 2014-11-06 08:51:24 boot alpine in VM… blacklist these modules in alpine or try to delete them from the kernel lib 2014-11-06 08:51:48 damm 2014-11-06 08:52:00 hute: can you add kernel params at boot? 2014-11-06 08:53:36 hute: ps… there was a command line in ubuntu 11.04 that caused a blank screen. did not power on lcd…. ill try to dig it up 2014-11-06 08:55:45 hute: yes i can. it is just the default iso from http://wiki.alpinelinux.org/cgi-bin/dl.cgi/v3.0/releases/x86_64/alpine-3.0.6-x86_64.iso so i git tab at "boot:" and start typing 2014-11-06 08:55:59 it boots using syslinux, since i used the unetbootin monstrosity 2014-11-06 08:56:38 hute: backlight is off i think…. https://wiki.archlinux.org/index.php/backlight 2014-11-06 08:56:46 hute: since its blank and working 2014-11-06 08:58:05 well I got to get away from a screen 2014-11-06 08:58:32 hute: definitely let me know about your experience with alpine and installing 2014-11-06 08:58:36 so we can make a good guide 2014-11-06 08:59:07 my network is shotty so no gaurentee i'll stay online, still futzing with my ZNC server 2014-11-06 09:00:10 api984: that is a good idea. ill try the backlight workarounds tonight, though i did fail to mention so far that the screen goes blank as in "display detects no signal coming out from the hdmi (nor vga) output" 2014-11-06 09:19:15 hute: might be something else… not sure … you need to try …. as long as machine is booted completely 2014-11-06 09:22:37 api984: i think it might be this, whatever it is, but none of those suggested fixes worked https://wiki.archlinux.org/index.php/Beginners%27_guide#Troubleshooting_boot_problems https://wiki.archlinux.org/index.php/Intel#Blank_screen_during_boot.2C_when_.22Loading_modules.22 2014-11-06 09:23:23 also i thought the fact that there is an open source driver now, would have fixed it, but there might be a conflict and i need to blacklist something, which i dont know the name of 2014-11-06 09:23:40 use Driver "vesa" not Driver "intel" 2014-11-06 09:23:49 to test if it works 2014-11-06 09:40:50 api984: that is for X, right? Im only planning to use virtual terminals, since it is intended to be a router/firewall 2014-11-06 09:41:22 hute: jup…. oops…. 2014-11-06 09:41:47 api984: seem to be wondering in wrong direction…. 2014-11-06 09:41:56 :D 2014-11-06 09:43:12 ill try some more desperate blacklisting tonight and try to figure out what ubuntu does and then report back if it works 2014-11-06 09:54:49 hute: thats cool 2014-11-06 09:55:00 hute: ill be here tomorrow ofcourse… 2014-11-06 10:03:46 hii 2014-11-06 14:50:18 hello, can you please help me ... is there any chance to get UFW in alpine ? 2014-11-06 14:50:31 (uncomplicated firewall) 2014-11-06 14:52:12 hi 2014-11-06 14:52:32 SVKroderik: we can probably add it but it will not be this week or next 2014-11-06 14:52:40 we are busy with wrapping up v3.1 release 2014-11-06 14:52:53 it will be great think (at least for me ) thnaks 2014-11-06 14:53:06 <^7heo> SVKroderik: do you have a link to the repo? 2014-11-06 14:53:07 *thing 2014-11-06 14:53:57 ^7heo" what repo ? 2014-11-06 14:56:12 <^7heo> nvm 2014-11-06 15:03:37 ^7heo: is this what you were looking for? https://code.launchpad.net/ufw 2014-11-06 15:03:45 ^7heo, which c99 features does gcc not support with -std=c99? 2014-11-06 15:04:24 <^7heo> ginjachris: probably. 2014-11-06 15:04:52 <^7heo> ginjachris: thanks for saving me a web search... but I expected SVKroderik to save us all the effort and simply share the link... 2014-11-06 15:05:05 <^7heo> dalias: I don't have the context anymore, sorry. 2014-11-06 15:05:46 ^7heo: it took me a while, kept finding front-ends for it.....I think that is the right one 2014-11-06 15:05:47 aaa taht repo, you talking about https://launchpad.net/ufw/+download 2014-11-06 15:05:49 sorry 2014-11-06 15:06:01 it was beginning of the autoconf conversatio 2014-11-06 15:06:06 https://launchpad.net/ufw/ 2014-11-06 15:06:41 <^7heo> dalias: yeah, I remember. I meant "I don't have the technical context" (i.e. lost the stdout/stderr) 2014-11-06 15:06:44 <^7heo> so I don't remember. 2014-11-06 15:07:10 ah :/ 2014-11-06 15:07:11 ok 2014-11-06 15:08:49 quick one: siege seems to unexpectedly exit or more commonly segfault for me...can anyone else replicate this? 2014-11-06 15:09:43 this is on 64 bit (not tested 32 bit) and 3.x release 2014-11-06 15:11:31 doing a '-g' (single get) is fine 2014-11-06 15:11:54 but it seems to die if I try and do a proper test after a handful of requests 2014-11-06 15:13:05 aaah interesting.....verbose mode seems to break it 2014-11-06 15:13:21 it's fine in quite mode (add -q flag) 2014-11-06 15:13:51 lol i spoke too soon 2014-11-06 15:14:43 it's better but still segfaults 2014-11-06 15:14:51 google.com kills it 2014-11-06 15:15:02 google.co.uk doesn't 2014-11-06 15:15:08 redirected related? 2014-11-06 15:15:21 this will need a smarter man than I to fix! 2014-11-06 15:29:44 <^7heo> ginjachris: it's a google repellent. Try with searchpage. 2014-11-06 15:29:46 <^7heo> ACTION hides 2014-11-06 15:30:10 <^7heo> ginjachris: what's the package? 2014-11-06 15:36:08 <^7heo> ginjachris: siege@testing ? 2014-11-06 15:39:56 sorry I had to step away 2014-11-06 15:40:05 errr yeah it's probably in testing 2014-11-06 15:40:25 yeah must be 2014-11-06 15:40:32 <^7heo> okay 2014-11-06 15:41:07 sometimes it will work (even google.com) but others it segfaults almost immediately 2014-11-06 15:41:59 it seems generally more reliable in quiet mode 2014-11-06 15:44:23 even if it were, say, related to a lack of available sockets (ports) it shouldn't seg fault so far as I know 2014-11-06 15:44:33 <^7heo> it seems to have more than one problem. 2014-11-06 15:44:41 <^7heo> I've spotted a free segfault. 2014-11-06 15:44:54 free! everyone loves free stuff ;) 2014-11-06 15:45:28 <^7heo> __stack_chk_fail () at src/env/__stack_chk_fail.c:18 2014-11-06 15:45:47 <^7heo> and now it hangs 2014-11-06 15:45:56 wow cool 2014-11-06 15:46:08 how does one troubleshoot these things? Valgrind? 2014-11-06 15:46:20 <^7heo> this code seems to be sucky 2014-11-06 15:47:42 :( sucky and code are not good together 2014-11-06 15:49:38 <^7heo> well, you've seen that, haven't you? 2014-11-06 15:50:09 indeed 2014-11-06 15:50:33 but I'm no coder 2014-11-06 15:50:42 just reasonably logical 2014-11-06 15:50:51 <^7heo> almost nobody is a coder. 2014-11-06 15:56:16 I think it's time for a cup of tea 2014-11-06 15:56:25 then a bug report on siege 2014-11-06 16:07:50 ok bug #3501 submitted 2014-11-06 16:15:16 gotta go soon so I shall leave you with some DnB: http://www.youtube.com/watch?v=pdYg7FerugE 2014-11-06 16:21:47 and one more for good luck: http://www.youtube.com/watch?v=6vj6O3ITFIA 2014-11-06 16:23:53 bye all 2014-11-06 16:31:36 <^7heo> first one is "unavailable", thanks GEMA... 2014-11-06 16:31:42 <^7heo> GEMA-hole. 2014-11-06 21:16:22 hute: any luck? 2014-11-06 21:49:12 SVKroderik, ncopa: ufw is in testing since a long time 2014-11-06 21:49:28 0.33 version 2014-11-06 21:49:52 thanks to barthalion 2014-11-06 21:52:23 I will make a package for a gui for ufw: http://gufw.org/ 2014-11-06 22:59:51 hmm, actually ufw is broken... 2014-11-06 23:01:04 we need 0.34rc version 2014-11-06 23:01:18 and ip6tables as dependency 2014-11-07 00:06:50 Hello, I would need some advices running alpine inside lxc. I choosed to do the bridge nat configuration, which works fine. However it's impossible to resolve any dns from insite the alpine lxc guest. There is apparently some problems in the alpine.in lxc template script which stop OpenRC at "Starting busybox syslog ... [ ok ]". I think making a screenshot is adequat for showing the situation: http://pix.toile-libre.org/upload/original/14153187 2014-11-07 00:07:03 inside * 2014-11-07 00:11:11 Goodbox: I haven't touched lxc yet 2014-11-07 00:11:30 lemme try 2014-11-07 00:11:36 I have been meaning to get to it 2014-11-07 00:12:26 :o don't waste your time, i'm juste taking a lxc try 2014-11-07 00:13:29 Goodbox: oh no problem 2014-11-07 00:13:37 I am needing to learn about it for the server I am setting up 2014-11-07 00:14:01 the strange thing is that the guest is able to ping my dns box, but can't resolve names 2014-11-07 00:14:29 can you bing ot the internet? 2014-11-07 00:14:37 *ping 2014-11-07 00:14:42 yes 2014-11-07 00:14:46 oh deard god no bing bleck :p 2014-11-07 00:14:50 try running the command 2014-11-07 00:14:57 dig @8.8.8.8 google.com 2014-11-07 00:15:07 you may have to install dig 2014-11-07 00:17:07 yeah, can't actually 2014-11-07 00:18:13 Since OpenRC fails, we just can lauch a shell, but networking and dependencies servires stills down 2014-11-07 00:20:02 services * ^^" 2014-11-07 00:30:02 setting up my virtual bridge 2014-11-07 00:30:05 one sec 2014-11-07 00:30:44 ow, thanks you systmkor3 :) 2014-11-07 00:32:03 Goodbox: I am systmkor but my fricking network her is shotty so i get kicked and then auto-reconnect and get bumped up to systmkor2 then systmkor3 etc 2014-11-07 00:32:05 :P 2014-11-07 00:32:55 haha ok, 2014-11-07 00:33:06 i just tested the bridge with a debian template 2014-11-07 00:33:36 same problem with network, but wheezy's init starts well, and shows the login program 2014-11-07 00:33:43 hmm arf 2014-11-07 00:37:09 i try a static configuration for the guests 2014-11-07 00:39:05 usually if you can't get dns records that means 2014-11-07 00:39:24 you aren't accessing your dns server properly 2014-11-07 00:39:37 and possibly may mean that your dhcp server is screwy 2014-11-07 00:41:35 yeah, this is my older debian running dnsmasq 2014-11-07 00:42:12 hrmm looks like lxc is missing depencies 2014-11-07 00:42:39 since running lxc-top says 'lua no such file' 2014-11-07 00:45:58 effectively, i just read the APKBUILD which doesn't append lua 2014-11-07 00:46:23 but witg configure --enable-lua 2014-11-07 00:47:07 ah no, i'm wrong 2014-11-07 00:47:41 missed lua5.2-dev sorry 2014-11-07 00:49:51 says it is missing lua lfs package 2014-11-07 00:56:44 so missing dependencies i guess would be lua5.2-dev and lua5.2-lfs 2014-11-07 01:10:12 Goodbox: I mispoke it is under lua-filesystem 2014-11-07 01:14:30 Ok 2014-11-07 01:14:46 I created a normal bridge to test 2014-11-07 01:15:01 it works with debian as guest 2014-11-07 01:15:09 I try with alpine 2014-11-07 01:16:54 how are you starting up the container? 2014-11-07 01:17:12 I do lxc-start -nalp 2014-11-07 01:17:15 with alp as name 2014-11-07 01:18:46 by default he'll require to be root 2014-11-07 01:19:26 else he says the container doesn't exists 2014-11-07 01:20:04 yah when I did that it just seems to hang 2014-11-07 01:20:09 and I can't get access to a shell 2014-11-07 01:20:34 on "Starting busybox syslog…" ? 2014-11-07 01:20:59 try lxc-attach -n thename 2014-11-07 01:21:08 you should drop an ash 2014-11-07 01:21:24 an ash? 2014-11-07 01:21:38 the busybox's posix shel 2014-11-07 01:21:44 shell* 2014-11-07 01:22:49 i get lxc_container: Operation not permitted - failed to set namespace 'pid' 2014-11-07 01:22:49 aparently ash inherit the environment from the host 2014-11-07 01:23:03 ow strange 2014-11-07 01:25:14 did you use brctl to set up your bridge then turn that bridge device up with ip link set dev br0 up 2014-11-07 01:26:17 systmkor3: yeah, the same with ifconfig 2014-11-07 01:26:41 i just dropped the network in alpine container now 2014-11-07 01:27:37 by doing ifconfig eth0 netmask 255.255.255.0; ifconfig eth0 192.168.122.2; route add default gw 192.168.122.1 2014-11-07 01:27:43 inside the container 2014-11-07 01:28:29 do you have your bridge attached to your interface 2014-11-07 01:28:36 that then talks to your dnsmasq server? 2014-11-07 01:28:48 if so you probably can just do dhcp 2014-11-07 01:28:50 yeah 2014-11-07 01:30:10 but OpenRC fails anyway 2014-11-07 01:34:30 hrm :\ 2014-11-07 01:41:22 all right, thanks you anyway :) 2014-11-07 01:41:52 yah I don't know 2014-11-07 01:41:55 for the moment I found lxc more simpler than openvz and vzctl 2014-11-07 01:42:04 yahhh... 2014-11-07 01:42:41 I would say docker is the most friendly but it is a bit neadlessly bloated at the moment 2014-11-07 01:42:48 and is missing some of the kernel security features that lxc has 2014-11-07 01:44:28 yes, another point interresting is the lxc's api ported to various languages, dunno how it's for openvz 2014-11-07 01:48:05 yah very nice 2014-11-07 01:48:12 I think docker has that as well but I don't remember 2014-11-07 01:48:36 also docker is getting some decent proof by being integrated into CoreOS as apart of service deployment 2014-11-07 01:50:58 so it would be nice if they worked to gether or complemented each other 2014-11-07 01:51:16 like lxc became strictly the container tech and docker became the management system ontop of lxc 2014-11-07 01:51:21 for those with more enterprise needs 2014-11-07 01:51:45 yeah i see 2014-11-07 01:51:54 with appliances frontends etc 2014-11-07 01:52:32 certainly nagios integrations ^^' 2014-11-07 01:55:37 since clients are linux-compatible environments, i really doesn't see why to deploy citrix or XCP solutions 2014-11-07 01:56:04 that's really hard to maintain, and so much heavy compared to LXC as example 2014-11-07 01:56:08 well KVM or Citrix virtualization 2014-11-07 01:56:15 solves a subtley different problem 2014-11-07 01:57:02 yeah, the hardware virtualisation and ressources repartitions ? 2014-11-07 01:57:10 yah 2014-11-07 01:57:35 I think isn't the need of small buisness 2014-11-07 01:57:44 it would be nice to see xen almost like a thin-hypervisor with tons of distributed computation support 2014-11-07 01:58:01 Goodbox: yah for a small business Xen isn't needed as much 2014-11-07 01:58:07 unless you need to do some Windows servers and such 2014-11-07 02:00:12 I recently tested the last gnome-shell, they have make a nice and smart frontent to qemu/kvm and spice 2014-11-07 02:00:52 I would like to see lxc become a default install means for main attack vectors 2014-11-07 02:00:59 of laptops and servers 2014-11-07 02:01:06 yeah me too 2014-11-07 02:01:06 so like Firefox, Chrome, PDF viewers 2014-11-07 02:01:33 one sort of wayland/xorg intergration of lxc containers 2014-11-07 02:01:38 so I am hoping we can add some kind of support in Alpine for making it easer to get an application that you want but in an lxccontainer 2014-11-07 02:01:42 with reasonable defaults 2014-11-07 02:02:04 yah I can't wait for wayland 2014-11-07 02:02:12 hahah :p 2014-11-07 02:02:17 same 2014-11-07 02:03:11 they're probably discussing about changing the world 2014-11-07 02:03:18 wayland guys 2014-11-07 02:05:18 I wonder how many time we would have a X11 running near wayland to run the thousands of xlib/xcb programms 2014-11-07 02:05:27 -m 2014-11-07 02:07:43 oh i don't know it will change the world 2014-11-07 02:07:50 i just like the base architecture better 2014-11-07 02:08:06 not nearly as poorly designed as x server 2014-11-07 02:08:42 yeah in fact i didn't find a similar expression of "refaire le monde" in english 2014-11-07 02:09:11 it's when folks are talking together about what's would be better 2014-11-07 02:09:44 now if only Alpine could have the man power of OpenBSD :P 2014-11-07 02:09:52 haha :D 2014-11-07 02:09:53 maybe minus some of their inane ways 2014-11-07 02:11:19 I prefer have alpine's man and process confinement, cpu taskset and of courses, the fine-grained smp of linux than the damn slow scheduler of OpenBSD 2014-11-07 02:12:18 in fact, I believe that obsd isn't viable at all for 2 or more cpu 2014-11-07 02:12:25 ACTION -> [] 2014-11-07 02:13:39 but I admire the clarity and homogeneity, i love alpine for some good obsd inspirations 2014-11-07 02:14:21 well yah that is more of what I was referring to 2014-11-07 02:14:34 without "nginx in default install with apache welcome page" 2014-11-07 02:15:01 i dunno if it's a troll, but obsd guys should update the default http server page of new installs ^^ 2014-11-07 02:16:49 globally, i found obsd as a well achieved utopic OS 2014-11-07 02:20:27 hopefully that can somewhat happen here 2014-11-07 02:23:02 to me Alpine brings true innovations to the cahotic linux world 2014-11-07 02:23:57 I ported many of programs I use, but i miss some experiences and time as maintainer 2014-11-07 02:24:32 I hope be able to maintain and host my few APKBUILDS one day 2014-11-07 02:28:04 chaotic* 2014-11-07 02:29:24 yah 2014-11-07 02:29:38 only thing I don't like about Alpine is yet another package manager 2014-11-07 02:29:55 i want to see if I can at least personally work in Nix package manager 2014-11-07 02:30:04 in some stable fashion 2014-11-07 02:30:16 since it looks like the first sensible package manager 2014-11-07 02:30:31 my beaf with it now is that it uses waayyy to much perl. 2014-11-07 02:30:37 I may try to port it to go :P 2014-11-07 02:30:53 ahhh i understand 2014-11-07 02:31:16 the division of work is a society bug|feature 2014-11-07 02:32:39 yeah, perl is a bit unreadable for non-sadic humans 2014-11-07 02:33:23 I personnaly adhere the way that package management is handled by a posix shell 2014-11-07 02:33:37 adere, i'm not sure 2014-11-07 02:34:35 adhere*… didn't learned english at shool, but's that's my favourite human way to talk simply 2014-11-07 02:38:56 like if, by a certain heredity it physically became the right way to exprime via this sufficiently rich and concise tool 2014-11-07 02:39:18 by analogy to the performance/security aspects ofoperating systems 2014-11-07 02:42:12 i found we're so young with the ability of computers ressources 2014-11-07 02:42:45 capabilities * 2014-11-07 02:43:32 ACTION hears "The Dark Side of the Moon - Pink Floyd" 2014-11-07 02:44:21 Goodbox: the problem with the simple POSIX package management 2014-11-07 02:44:55 is that its non-deterministic, has unecessary state, and can result in an undefined state due to race conditions 2014-11-07 02:45:41 for example if I start updating my with POSIX style package manager, and my computer crashes in the middle of a critical seciton such as replacing your kernel image 2014-11-07 02:45:48 you are hosed 2014-11-07 02:45:55 from an embedded systems standpoint 2014-11-07 02:46:21 basically I think the goals and features of Nix package manager is what all package managers at a minimum should have 2014-11-07 02:46:26 though I don't know about it's implementation 2014-11-07 02:47:22 i see, i like to believe that good programmers have good ideas :p 2014-11-07 02:48:44 in a certain way, a shell might rest the user administrator interface 2014-11-07 02:48:48 openssl :3 2014-11-07 02:49:03 Diftraku: ?_? 2014-11-07 02:49:11 he does a nice shot 2014-11-07 02:49:14 ACTION wonders what "posix package management" is 2014-11-07 02:49:14 :D 2014-11-07 02:49:38 i was talking the way to handle the package management 2014-11-07 02:49:49 via a posix shell 2014-11-07 02:50:02 dalias: i don't think there is an actual POSIX package management 2014-11-07 02:50:11 but common style amongst existing POSIX systems 2014-11-07 02:50:39 i also don't see how replacing your kernel should hose anything 2014-11-07 02:51:05 you make the switch via atomic rename() of either the kernel file or a config file after all the other stuff is in place 2014-11-07 02:51:24 as long as the filesystem is safe against crashes, the upgrade is safe 2014-11-07 02:53:53 my first alpine installation succeed, she never crashed since following edge 2014-11-07 02:54:28 this is a first in my little nixer life 2014-11-07 03:12:19 dalias: well renaming one file is atomic 2014-11-07 03:12:48 and by hosed I just mean you are now in an unknown state 2014-11-07 03:13:08 dalias: so for example if your kernel is renamed but then your computer crashes before the initramfs 2014-11-07 03:13:10 is renamed 2014-11-07 03:13:28 now you have a new kernel with an old initramfs 2014-11-07 03:13:43 also by using nix's style of package management 2014-11-07 03:14:06 you can have multiples of the same application but different versions with zero confclits 2014-11-07 03:14:29 you can basically do everything that is done by nix 2014-11-07 03:14:47 dalias: but in its totality it creates a more complex system 2014-11-07 03:15:07 systmkor3, ideally the initramfs wouldn't exist :) 2014-11-07 03:15:28 dalias: sure, however I assume you are getting my point though 2014-11-07 03:16:24 no, i'm not. your point is about atomic version switching which is rather orthogonal to nix's declarative, reproducable package generation and installation 2014-11-07 03:16:41 either can be achieved without the other 2014-11-07 03:17:05 dalias: it's not orthogonal 2014-11-07 03:19:03 well nix more specifically uses a functional & lazzy language for expressing an install 2014-11-07 03:19:27 dalias: also rename() isn't architectually atomic 2014-11-07 03:19:45 your OS could crash mid rename operation 2014-11-07 03:20:49 and one of nix's goal is for atomic package/dependency switching 2014-11-07 03:21:32 dalias: so I guess I am not understanding your point 2014-11-07 03:23:22 since a reproducable package install, and upgrade needs to be an atomic operation 2014-11-07 03:26:01 rename is atomic for a non-crashing system 2014-11-07 03:26:32 whether it's atomic under crashes is a matter of your filesystem and even your storage device (some storage devices horribly corrupt data under power loss, for example) 2014-11-07 03:27:05 but nix cannot fix a filesystem or storage device that's unsafe against crashes 2014-11-07 03:27:24 well yes 2014-11-07 03:27:28 but if there is a system crash 2014-11-07 03:27:35 then you don't end up in an undefined state 2014-11-07 03:27:39 yes you do 2014-11-07 03:27:43 which as far as I know is the point 2014-11-07 03:27:52 no... 2014-11-07 03:27:53 nix has no say in this 2014-11-07 03:28:03 after hardware failure the whole machine state is undefined 2014-11-07 03:28:05 by undefined state I mean the packages 2014-11-07 03:28:08 it can't fix this 2014-11-07 03:28:11 no the actual machine state itself 2014-11-07 03:28:30 that includes the packages 2014-11-07 03:28:40 they could suffer from arbitrary corruption during crash 2014-11-07 03:28:41 that is waht I mean, an undefined state with respect to package management 2014-11-07 03:28:53 some ssds are known to badly corrupt data if power is removed during a write 2014-11-07 03:28:56 if the hardware fubars during and upgrade, yoi got more things to worry than what state the os is in 2014-11-07 03:29:12 also, backup your shit 2014-11-07 03:29:18 well yah 2014-11-07 03:29:25 nix solves important problems 2014-11-07 03:29:34 what happens under crashes is NOT one of the problems it solves 2014-11-07 03:29:36 they're unrelated 2014-11-07 03:30:07 so your saying a system crash is unrelated to package management 2014-11-07 03:30:14 yes 2014-11-07 03:30:18 if you want to protect against hand of gos, get N+1 IBM Z-series mainframes 2014-11-07 03:30:28 Diftraku: I am not saying hand of gods 2014-11-07 03:30:35 I am saying for example a power outage 2014-11-07 03:30:47 those fuckers can hotswap a cpu 2014-11-07 03:30:49 if you want to protect against power outage get a $30 ups 2014-11-07 03:30:57 ^ 2014-11-07 03:30:57 which would be par for the course of embedded devices 2014-11-07 03:31:37 dalias: agreed if you have a catastrofic clusterfuck on your hardware it doesn't matter 2014-11-07 03:31:41 a 5€ embedded kit needs to be as static as possible if power is not reliable 2014-11-07 03:32:00 okay price is arbitrary 2014-11-07 03:32:01 for an embedded device, that probably means a few big capacitors -- enough to last a few hundred ms while you sync the flash drive and remount ro 2014-11-07 03:32:04 like openwrt on a router 2014-11-07 03:32:10 and teh embedded device has to be upgraded at some point 2014-11-07 03:32:27 dalias: while I agree that is better practice for hardware/system design 2014-11-07 03:32:33 yes, but upgrades should take place under controlled environment 2014-11-07 03:32:42 you can achieve partial similarity with nix 2014-11-07 03:32:47 Diftraku: not all upgrades can be 2014-11-07 03:32:50 again embedded 2014-11-07 03:32:57 you don't upgrade a running system during rush-hour 2014-11-07 03:33:05 *sigh* 2014-11-07 03:33:06 I didn't say you should 2014-11-07 03:33:11 then you make sure you got a backup on hand 2014-11-07 03:33:35 okay so an embedded situation problem, is that let's say I have a computer on the moon 2014-11-07 03:33:37 *nix doesn't prevent forest fires, only you can 2014-11-07 03:33:47 I didn't say nix prevents forest fires 2014-11-07 03:34:02 i am saying it reduces for package management undefined state 2014-11-07 03:34:07 within reason if there is a system crash 2014-11-07 03:34:08 applies to hardware failures and etc 2014-11-07 03:34:12 not the system blowing up 2014-11-07 03:34:17 because then you wouldn't have a system 2014-11-07 03:34:24 systmkor3, then you have two OS's, one that's on rom and never changed, that's used as a maintenance mode if something goes wrong upgrading the other 2014-11-07 03:34:53 dalias: yes agreed 2014-11-07 03:34:56 dalias: or better yet, the rom loads the os off non-volatile and defaults to maintenance 2014-11-07 03:34:58 and some kind of watchdog timer to switch to the maintenance one 2014-11-07 03:35:04 but you usually would prefer the rom as a ultimate fallback 2014-11-07 03:35:25 dalias: again you can accomplish what nix does 2014-11-07 03:35:51 but you would need redundant partitions, file system overlay, and still other dependency hell issues to deal with 2014-11-07 03:35:58 no 2014-11-07 03:36:07 no to what 2014-11-07 03:37:45 what nix accomplishes is predictable package builds and ability to rollback, have multiple versions installed and not interfere with each other, etc. 2014-11-07 03:39:55 whether you want/need that is independent of how you handle robustness of upgrades 2014-11-07 03:40:09 yes i <3 <3 <3 anything declarative 2014-11-07 03:40:24 but declarative is not a prerequisite for robust upgrades 2014-11-07 03:40:32 I still think you are wrong but I will go back and re-read material before I continue, and change mind accordingly 2014-11-07 03:40:33 nor is it sufficient 2014-11-07 03:43:44 dalias: explain your definition of 'robust upgrade' 2014-11-07 03:44:05 just to make sure we aren't talking past each other 2014-11-07 03:45:22 there are two we could look at. one is upgrading without a race window where the program is broken due to a mix of old and new files. this has nothing to do with system crashes 2014-11-07 03:45:54 the second would be extending that to what happens when the system goes down asynchronously during the upgrade process 2014-11-07 03:47:15 and you aren't euphimistcally meaning 'goes down asynchronously' as a system crash 2014-11-07 03:47:34 since there 'system crash' is broader 2014-11-07 03:47:41 we could break this down into multiple levels too 2014-11-07 03:48:02 e.g. kill -9 -1 2014-11-07 03:49:04 power button 2014-11-07 03:49:05 unplugging 2014-11-07 03:49:09 lightning strikes machine 2014-11-07 03:49:10 etc. 2014-11-07 03:50:48 and your claim is that nix can't recover/end up and undefined package management state if there is a unplugging 2014-11-07 03:51:37 whether it can or can't depends on whether (1) the filesystem, and (2) the hdd/ssd/whatever, can survive unplugging 2014-11-07 03:52:06 conceptually you have no guarantee that the hdd contains anything but random bytes 2014-11-07 03:52:20 you do have some conceptually 2014-11-07 03:52:29 doepends on ho you want to conceptualize it 2014-11-07 03:52:35 not without further guarantees from the device and filesystem 2014-11-07 03:53:06 if you have a hdd that's designed to survive losing power, and ext3 or ext4 with journal, then you're in pretty good shape 2014-11-07 03:53:20 so it depends here 2014-11-07 03:53:26 are you assuming that during a power outage 2014-11-07 03:53:33 that old packages 2014-11-07 03:53:35 if you're using ntfs on a compressed and encrypted volume.... 2014-11-07 03:53:38 could be damaged 2014-11-07 03:53:53 yes, anything could 2014-11-07 03:54:02 well the cpu you could have blown up 2014-11-07 03:54:24 for example accesses to the directories old stuff is in will involve atime updates 2014-11-07 03:54:30 so you are assuming nothing i static 2014-11-07 03:54:34 which will cause writes to those directory tables 2014-11-07 03:54:47 for the given non volatile storage and file system 2014-11-07 03:55:21 *is static 2014-11-07 03:56:04 that is to say file that I don't touch are assumed to be corrupted during this power outage 2014-11-07 03:56:10 *files that I 2014-11-07 03:59:10 not necessarily. just that you can't assume they're non-corrupted either _unless_ you have hardware and a filesystem type that attempts to make such promises 2014-11-07 03:59:33 so yah I would agree nix wouldn't be sufficient 2014-11-07 03:59:41 with most decent desktop/server grade storage devices and a decent fs like ext3/4 you're probably good 2014-11-07 03:59:51 you would need so error correcting properties below the package manager 2014-11-07 04:00:15 *such error 2014-11-07 04:00:30 this is why i say it's not sufficient (you need other layers) 2014-11-07 04:00:37 oh agreed on that 2014-11-07 04:00:45 the reason i say it's not necessary is that it's easy, if you have these layers, to make other package systems safe 2014-11-07 04:01:01 that's were I would disagree 2014-11-07 04:01:14 e.g. first record and sync that you're in the middle of updating package X, then do the upgrade, then sync and record that it's finished 2014-11-07 04:01:34 because even if you have those underlying layers that doesn't guarentee a deterministic update 2014-11-07 04:01:36 and on boot, if you find that the first record was made, but not the second... 2014-11-07 04:01:47 ...you just repeat the package unpacking/install and continue 2014-11-07 04:02:05 well that's assuming non of the old packages were touched 2014-11-07 04:02:10 during a traditional package update 2014-11-07 04:02:23 which traditional package updates destroy old packages as you go 2014-11-07 04:03:48 so if you are updating package X 2014-11-07 04:04:30 but package Y depends on it and package Y is necessary in the chain of booting/and finishing the upgrade 2014-11-07 04:04:51 but there is system downed before it finished updating Y 2014-11-07 04:05:13 and the package Y (old) is incompatible with package X (new) 2014-11-07 04:05:21 then that method wouldn't work 2014-11-07 04:05:41 dalias: btw thank you for the discussion 2014-11-07 04:08:02 dalias: so just having a verified package updated one at a time doesn't guarentee the whole update will be recoverable if there is a failure mid-way 2014-11-07 04:08:56 indeed, if your init and package system are full of dependency hell then it will be a problem 2014-11-07 04:11:01 which is kinda par for the course 2014-11-07 04:11:04 on modern OS's 2014-11-07 04:12:32 ACTION glares at systemd :) 2014-11-07 04:13:49 ideally the path from init to 'check for interrupted package upgrade and resume' would involve nothing but a static linked, atomically-replacable busybox and script files 2014-11-07 04:14:08 plus the task of getting dispirate coding projects to maintain a pristine prevention of dependency hell 2014-11-07 04:14:33 is many orders of magnitude more difficult than just using something equivilent to nix 2014-11-07 04:16:43 well the it would be from the kernel through init to back to verifiying a package upgrade would have to be atomic 2014-11-07 04:18:28 so if all of that was one file or you do a do a rename of a root directory after writing multiple files and verifying them so that it if effectively atomic 2014-11-07 04:18:49 the individual components involved in this early startup would need to be safe (preferably, atomic) to upgrade individually, but not necessarily as a group 2014-11-07 04:20:23 well it would be necessary if there is a incompatibility 2014-11-07 04:20:29 between old and new components 2014-11-07 04:20:41 that's why there shouldn't be 2014-11-07 04:21:01 that seems arbitrary 2014-11-07 04:21:12 i would view that as a more fragile system 2014-11-07 04:21:41 because saying there shouldn't be any component version issues 2014-11-07 04:21:47 would apply to all of package management 2014-11-07 04:23:27 meaning I would say that initself isn't guaranteeable 2014-11-07 04:23:54 which is why I would argue that something like nix would be effectively necessary 2014-11-07 04:23:58 well i think they're different philosophies 2014-11-07 04:24:49 i would say that's a bit of a stretch but continue 2014-11-07 04:25:37 mine is that the core stuff needed to get the system up and running should have minimal dependencies, those dependencies should be documented, noncircular, and a matter of policy rather than chance, and the components used should all be designed for maximum reliability 2014-11-07 04:26:01 I would agree to that 2014-11-07 04:26:07 but even if that was a goal 2014-11-07 04:26:21 that doesn't guarentee it to the same degree I would argue nix would 2014-11-07 04:27:22 don't get me wrong. i really like the idea of nix. (altho i'm not a huge fan of the implementation; it relies on a lot of bloated and ugly components like perl, sqlite, etc.) 2014-11-07 04:27:38 full agreement there 2014-11-07 04:28:01 well for work i am trying to make our embedded devices use alpine 2014-11-07 04:28:13 but it would invaluable if I could get nix working in conjuction with alpine 2014-11-07 04:28:14 but i also know lots of ppl have an (imo unreasonable) aversion to declarative stuff and i don't think pushing them to adopt this kind of system design (unless they _want_ to) is a high priority 2014-11-07 04:28:21 :) 2014-11-07 04:28:25 so I want to rewrite nix in go 2014-11-07 04:28:49 why not in make? :) 2014-11-07 04:28:49 i think databases for that like sqlite should be an option but I don't know abut a dependency 2014-11-07 04:29:02 my guess is that you could get away with a key value store 2014-11-07 04:29:07 which I think would be reasonably minimal 2014-11-07 04:29:10 (then you already have a nice declarative language as the basis ;-) 2014-11-07 04:29:16 yuuuuup 2014-11-07 04:29:26 well fuck it 2014-11-07 04:29:34 I'll try to get started by the end of the week 2014-11-07 04:29:42 I haven't finished reading all the nix papers yet 2014-11-07 04:29:50 or had a long enough examination at the source code 2014-11-07 04:29:53 plus fuck perl 2014-11-07 04:30:51 iirc there's already one other declarative package system like nix in a less-ugly implementation 2014-11-07 04:30:59 but i don't remember what it's called 2014-11-07 04:31:49 well looking at the NixOS paper 2014-11-07 04:32:03 there goal is a declartive configuration in conjuction with package updates 2014-11-07 04:32:17 Nix + Alpine + Ansible 2014-11-07 04:32:25 I could rule the world! :P 2014-11-07 04:32:57 :) 2014-11-07 04:33:59 to bad that gobolinux died 2014-11-07 04:34:03 too* 2014-11-07 04:34:05 rah 2014-11-07 04:35:11 NixOS and nix PM are different from the guix project isn't ? 2014-11-07 04:36:21 uhhh 2014-11-07 04:36:40 i would assume so 2014-11-07 04:36:43 guix is based on nix 2014-11-07 04:37:05 guix uses guile as language 2014-11-07 04:37:53 to build packages 2014-11-07 04:38:09 ahhh fuck really 2014-11-07 04:38:13 ok, thanks 2014-11-07 04:38:31 new better way to do package management 2014-11-07 04:38:41 splinter code on spererate projects 2014-11-07 04:38:46 with almost identical features 2014-11-07 04:38:48 and implementations 2014-11-07 04:38:51 go 2014-11-07 04:39:41 grabbing food be back in a bit 2014-11-07 04:40:00 ah , and gobolinux is alive ;) 2014-11-07 04:42:35 I dunno what left me think that. The last time i installed its, there was no available mirrors up to date if I remember. 2014-11-07 04:44:55 well, it was dead four years 2014-11-07 04:45:43 it reborn six mounths ago 2014-11-07 04:45:49 ok, it explains it 2014-11-07 04:46:06 ACTION take a tour on the web 2014-11-07 04:46:45 cool, new things to test, thanks :p 2014-11-07 04:47:23 ;) 2014-11-07 04:54:21 out of subject, according to you, what would you use to integrate [para-]virtualized instances of X11 clients/sessions in your local desktop ? I know things like xforwarding, xephyr, vnc or spice protocols. 2014-11-07 04:56:20 One day I tried to get back a X client from a gentoo chroot in alpine… I'll probably will have better results with a lxc 2014-11-07 04:57:58 Goodbox: Goodbox I don't know but it's a good idea 2014-11-07 04:58:11 there is distro that does something like that fo ryou 2014-11-07 04:58:14 I don't remember the name 2014-11-07 04:58:41 Bedrock Linux? 2014-11-07 04:59:09 not what i was thinking 2014-11-07 04:59:20 ah, sorry 2014-11-07 05:00:23 but looks cool 2014-11-07 05:01:42 haha, long ago i read about a governement which assembled a backdoored graphical distro running KDE in chroot 2014-11-07 05:02:26 backdoored running KDE 2014-11-07 05:02:34 KDE is so poorly written it is a backdoor 2014-11-07 05:02:35 :P 2014-11-07 05:03:49 AmatCoder: it's Qubes 2014-11-07 05:03:51 https://wiki.qubes-os.org/ 2014-11-07 05:03:54 that is what I was thinking of 2014-11-07 05:04:13 oh, I see 2014-11-07 05:05:36 it kills d-bus problems 2014-11-07 05:07:36 it also would be nice to have a switch when installing a package 2014-11-07 05:08:05 that it's designed to run in a lxc and proper grsec policy 2014-11-07 05:08:22 maybe application not at the actual package granularity 2014-11-07 05:09:50 qubes seems using rpm on the dom0 2014-11-07 05:10:58 yah 2014-11-07 05:11:06 nothing groundbackingly new 2014-11-07 05:11:16 but I think designed to have to do less manual configuration 2014-11-07 05:11:25 aka less boilerplate 2014-11-07 05:13:02 I never tried to launch a xen hypervisor in qemu but i'm sure it should work 2014-11-07 05:14:47 they drop back x11 clients trough the virtual network ? 2014-11-07 05:15:22 I don't know :\ 2014-11-07 05:16:23 it must be expensive for the hardware 2014-11-07 05:17:17 yuuup 2014-11-07 05:19:08 that's why lxc is probaby the best compramise 2014-11-07 05:19:16 sure :D 2014-11-07 05:20:57 but problem is 2014-11-07 05:21:07 there is way too much manual configuration to get it to work 2014-11-07 05:21:22 e.g. it would be nice just to download like firefox-hardened 2014-11-07 05:21:32 which runs with appropiate cgroups inside a lxc 2014-11-07 05:23:02 yes, even for osx integrators, they prefer emulate a graphic layer like parallels desktop does 2014-11-07 05:23:25 it's most complicated for the hardware 2014-11-07 05:23:32 but simpler for the user 2014-11-07 05:25:03 a nice xclient forwarding with correct xhost settings is difficult to do for a non computer scientist 2014-11-07 05:26:46 yup 2014-11-07 05:29:15 ah in fact that is a way i was expecting http://unix.stackexchange.com/questions/18003/linux-lxc-deploying-images-with-tiniest-possible-x11 2014-11-07 05:30:43 once configured, we theorically just have to set DISPLAY 2014-11-07 05:33:32 Goodbox: do it. :P and make a Alpine wiki article 2014-11-07 05:33:38 detailed how to 2014-11-07 05:35:09 ok 2014-11-07 05:35:12 x) 2014-11-07 05:35:33 i'll try several ways 2014-11-07 05:36:04 i really think the simpler is binding the same devices 2014-11-07 05:37:25 what do you mean? 2014-11-07 05:38:34 it's maybe insecure 2014-11-07 05:38:49 well everything is insecure 2014-11-07 05:38:49 programs may be able to catch others informations from others programs sharing the same device abstraction 2014-11-07 05:38:52 but is it more secure 2014-11-07 05:39:18 and is the model generic enough that it could be refined with out massive re-architecture 2014-11-07 05:39:22 if so 2014-11-07 05:39:24 build, break, fix, repeat 2014-11-07 05:39:28 AGILE ahoi 2014-11-07 07:28:51 systmkor3: sadly none, but i did not have much time to look at the issue last night. i tested all additional boot parameters i could think of, like disabling some other outputs with video= , blacklisting modules, setting modeset=0 on modules etc. ill test more during the weekend. 2014-11-07 07:29:52 dang 2014-11-07 07:31:02 but really didnt have time to go over it systematically. still have to look more at how ubuntu manages to boot and try to make my own kernel images. also i realised i tested with grsec the whole time, instead of the vanilla kernel. dont know if there is much difference in the way modules are handeled. 2014-11-07 07:33:12 well, time for the daily grind 2014-11-07 07:34:09 hute: https://www.youtube.com/watch?v=-ECUtkv2qV8 2014-11-07 07:44:04 systmkor3: thanks :) 2014-11-07 08:45:13 http://www.debianfork.org/ 2014-11-07 08:45:39 one day people's choice of distro will start "with systemd or not?" 2014-11-07 08:47:05 yep, that's why I am looking into alpine, amazing distro if a few things I need work (like zfs and openvswitch) 2014-11-07 08:48:12 openvswitch is heavily used by alpine folk, so you should be fine there. zfs not so much. clandmeter uses it a lot i think 2014-11-07 08:49:12 ScrumpyJack: is openvswitch on edge or 3.1.0 2014-11-07 08:49:22 because as far as I can tell it's not in 3.0.6 2014-11-07 08:58:12 yes, it's in edge :) 2014-11-07 08:58:32 which is where i sit, coming from arch :) 2014-11-07 08:59:09 but arch uses systemd (and I can't be bothered to change it myself), so bye bye arch 2014-11-07 08:59:25 and hello alpinelinux :) 2014-11-07 09:12:50 Tried once with openvswitch and couldn't make it through. So I left and checking back again ,_) 2014-11-07 09:13:55 <^7heo> moin @ 2014-11-07 09:47:22 Configured openvswitch, but machine cannot access internet (works fine with eth0 dhcp). 2014-11-07 09:48:09 <^7heo> are you configuring eth0 for your machine and br0 for openvswitch? 2014-11-07 09:51:06 ^7heo: See here: http://imgur.com/BRkctxY 2014-11-07 09:52:14 I have also configured libvirtd to work with openvswitch and started the default libvirt after making necessary configs. 2014-11-07 09:52:21 *config changes 2014-11-07 09:54:42 <^7heo> moment 2014-11-07 09:54:48 <^7heo> I have to re-assign a few tickets 2014-11-07 09:57:18 <^7heo> omg 2014-11-07 09:57:40 <^7heo> zenny: really, can't you paste CODE instead of images?! 2014-11-07 09:58:01 <^7heo> anyway 2014-11-07 09:58:19 <^7heo> you're not supposed to use eth0 directly, even for the host, after you configure bridging. 2014-11-07 09:58:24 <^7heo> maybe that is your problem? 2014-11-07 10:00:11 ^7heo: About images, I am running alpine currently in a kvm instance in ubuntu and cannot access the alpine instance of kvm using ssh. So I just took a screenshot. 2014-11-07 10:00:51 <^7heo> I figured that out. nevertheless, that's no excuse for wasting bw and screwing readability... 2014-11-07 10:00:56 <^7heo> readibiility* 2014-11-07 10:01:02 <^7heo> rah, fuck/ 2014-11-07 10:01:25 <^7heo> you can type in the alpine instance, right? 2014-11-07 10:02:10 However, I have a similar setup on the ubuntu kvm and working perfectly for months. 2014-11-07 10:02:18 yep I have access to the alpine instance. 2014-11-07 10:04:38 <^7heo> does it have network? 2014-11-07 10:06:48 nope. 2014-11-07 10:08:34 <^7heo> Is that the problem why you are writing here? 2014-11-07 10:10:20 yep 2014-11-07 10:13:27 <^7heo> You should have explained better. 2014-11-07 10:13:43 <^7heo> < zenny> Configured openvswitch, but machine cannot access internet (works fine with eth0 dhcp). 2014-11-07 10:13:52 Trying to follow this: http://kashyapc.com/2013/07/13/configuring-libvirt-guests-with-an-open-vswitch-bridge/ and http://dtucker.co.uk/hack/installing-kvm-libvirt-openvswitch-on-fedora.html 2014-11-07 10:13:53 <^7heo> "machine" doesn't mean shit. 2014-11-07 10:14:06 <^7heo> don't follow stuff. 2014-11-07 10:14:15 with OVS, you need to "ovs-vsctl add-br bridge;ifconfig eth0 0.0.0.0 up;ovs-vsctl add-port bridge eth0;dhcp-client bridge" 2014-11-07 10:14:26 <^7heo> That's probably written by fat-eating apes who have no clue about why things work. 2014-11-07 10:15:13 Jean-Scotch: That has been already done in the /etc/network/interfaces (see http://imgur.com/BRkctxY) 2014-11-07 10:15:58 hello 2014-11-07 10:16:04 <^7heo> moin ginjachris 2014-11-07 10:16:29 hooray it's Friday! 2014-11-07 10:16:55 ^7theo: I have no offense against anyone. If you know of how to make the host machine to access the net, advise please. 2014-11-07 10:17:24 <^7heo> I told you. 2014-11-07 10:17:50 <^7heo> but it was to explain how things work: if you're having access to the net with eth0 2014-11-07 10:18:08 <^7heo> you can't have connectivity in your vm 2014-11-07 10:21:16 zenny: your VM has a tap which is added to the bridge just before launch? 2014-11-07 10:25:02 I do it with "qemu ... -net nic,vlan=0,model=virtio,macaddr=$MACA -net tap,vlan=0,script=/etc/openvswitch/ovs-ifup-bridge,downscript=/etc/openvswitch/ovs-ifdown-bridge,ifname=$NAME0 ... " 2014-11-07 10:30:06 Jean-Scotch: no tap before the launch of the bridge. 2014-11-07 10:31:16 I mean before the launch of the VM 2014-11-07 10:55:11 Jean-Scotch: None 2014-11-07 10:56:17 I think this may be the reason your VM doesn't have net access as it is supposed to use a tap connected to the OVS. no? 2014-11-07 10:58:00 sorry can't help with kvm. xen all the way here :) 2014-11-07 10:58:19 while eth0 is connected to the bridge, you are not supposed to use it directly anymore 2014-11-07 11:02:00 can you get it to work without openvswitch? 2014-11-07 11:02:56 with brctl 2014-11-07 11:03:35 brctl addbr br0 2014-11-07 11:03:39 brctl setfd br0 0 2014-11-07 11:03:44 brctl addif br0 dummy0 2014-11-07 11:03:49 ip add add 192.168.1.1/24 dev br0 2014-11-07 11:03:53 ip link set dev br0 up 2014-11-07 11:03:56 and then 2014-11-07 11:09:18 (sorry, had to look up the kvm stuff 2014-11-07 11:09:37 http://www.linux-kvm.org/page/Networking 2014-11-07 11:09:48 section "You need a qemu-ifup script containing the following:" 2014-11-07 11:16:46 may i be bold and recommend alpine-xen for this? you'll find it extremely tidy, with alpine providing a perfect dom0 for your qemu+xen vms needs :0 2014-11-07 11:23:46 <^7heo> ScrumpyJack: you would be right, but he's using ubuntu + kvm as a host. 2014-11-07 11:24:18 <^7heo> and replacing kvm by Xen is one thing, but ubuntu by alpine might be... difficult for some people. 2014-11-07 11:32:41 ^7heo: This problem does not exist with other distro in kvm inside kvm (take debian). This has been an outstanding issue the last time I tested with alpine as well as this time (filed a bug report earlier). All others like slackware, centos works fine except alpine. BTW, I am trying to test alpine on a ubuntu+kvm+openvswitch+libvirtd machine, in production I usuually use either slackware, debian or centos. alpine seems closer 2014-11-07 11:33:32 <^7heo> =/ 2014-11-07 11:34:13 <^7heo> what can I say now... 2014-11-07 11:34:55 i'm confused. was alpine the guest or the host? 2014-11-07 11:36:31 <^7heo> he doesn't explain well, to say the least. 2014-11-07 11:36:50 <^7heo> I don't really know either, but I assume that he has problems with ubuntu + kvm hosting alpine as a VM. 2014-11-07 11:37:08 <^7heo> first thing I know, he said he had a problem with a "machine" 2014-11-07 11:38:13 oh well, back to work 2014-11-07 11:38:17 <^7heo> yeah 2014-11-07 11:58:00 right, bored now. gonna see if i can get X working in alpine 2014-11-07 12:05:20 <^7heo> it's very easy to do. 2014-11-07 12:05:30 <^7heo> setup-xorg-base 2014-11-07 12:06:24 i want to use twm 2014-11-07 12:07:45 <^7heo> twm? 2014-11-07 12:07:56 yup 2014-11-07 12:07:59 <^7heo> I use dwm 2014-11-07 12:08:31 can you try twm? it could be the package i made is broken :( it's in testing 2014-11-07 12:08:50 <^7heo> hmm 2014-11-07 12:09:23 <^7heo> I have to exit my current wm for that 2014-11-07 12:09:24 <^7heo> moment 2014-11-07 12:09:55 that's what xephyr is for 2014-11-07 12:11:19 twm doesn't work for me, but it could be just my X config is borked 2014-11-07 12:11:26 as nothing works :) 2014-11-07 12:11:38 <^7heo> okay 2014-11-07 12:11:40 <^7heo> moment 2014-11-07 12:11:42 if someone can test twm, at least i know what to rule out 2014-11-07 12:12:37 <^7heo> it works. 2014-11-07 12:12:43 <^7heo> now tell me how I can use it v_v 2014-11-07 12:12:50 <^7heo> how do I resize a window? 2014-11-07 12:12:56 <^7heo> do I have many workspaces? 2014-11-07 12:13:35 <^7heo> okay I found a way to resize the window vertically, but not horizontally 2014-11-07 12:13:51 yeah, you'll need to set all that up in a .twmrc as you want it 2014-11-07 12:13:58 i can't remember the defaults 2014-11-07 12:14:03 <^7heo> hmm 2014-11-07 12:14:11 <^7heo> do you know how to exit it? 2014-11-07 12:14:33 thanks for trying it. i wouldn't recommend using it if you already have a nice window manager :) 2014-11-07 12:14:34 <^7heo> oh I just found it 2014-11-07 12:14:36 <^7heo> desk menu 2014-11-07 12:14:47 <^7heo> nah but it's fun to tr 2014-11-07 12:14:48 <^7heo> try 2014-11-07 12:15:40 <^7heo> anyway, brb, changing wm 2014-11-07 12:15:54 i love it because it's very old, very small, it does what i want and i have simple needs :) 2014-11-07 12:16:31 <^7heo> back 2014-11-07 12:16:40 wb 2014-11-07 12:16:45 the greatest thing about twm is that it creates windows the right way 2014-11-07 12:17:06 that is to say, *i* create windows 2014-11-07 12:17:29 <^7heo> $ ls -lh $(whereis twm | cut -d\ -f2) 2014-11-07 12:17:30 <^7heo> -rwxr-xr-x 1 root root 171.6K Jun 23 10:24 /usr/bin/twm 2014-11-07 12:17:36 <^7heo> $ ls -lh $(whereis dwm | cut -d\ -f2) 2014-11-07 12:17:36 <^7heo> -rwxr-xr-x 1 root root 52.1K Aug 22 07:45 /usr/local/bin/dwm 2014-11-07 12:17:38 <^7heo> Just saying. 2014-11-07 12:18:23 <^7heo> then after I don't really know what you call "the right way of creating windows" :D 2014-11-07 12:18:54 you draw the rectangle 2014-11-07 12:19:08 by the way, there's which(1) 2014-11-07 12:19:27 <^7heo> not here. 2014-11-07 12:19:33 <^7heo> oh 2014-11-07 12:19:36 <^7heo> now it is. 2014-11-07 12:19:38 <^7heo> funny. 2014-11-07 12:19:51 <^7heo> new version of busybox I presume? 2014-11-07 12:19:52 hmm, nothing should be in /usr/local 2014-11-07 12:20:04 <^7heo> ScrumpyJack: of course stuff should be there. 2014-11-07 12:20:11 <^7heo> dwm is compiled by myself. 2014-11-07 12:20:19 <^7heo> so ofc it SHALL be there. 2014-11-07 12:20:34 i think abuild discourages stuff in /usr/local 2014-11-07 12:20:41 <^7heo> abuild yes 2014-11-07 12:20:47 <^7heo> make install no. 2014-11-07 12:20:59 <^7heo> and my way of using dwm is just: 2014-11-07 12:21:08 share your work - make a package :) 2014-11-07 12:21:16 <^7heo> git clone git://git.suckless.org/dwm && cd dwm && make && sudo make install 2014-11-07 12:21:19 <^7heo> done. 2014-11-07 12:21:27 <^7heo> dwm is about NOT using a package manager. 2014-11-07 12:21:35 yes i remember now 2014-11-07 12:21:35 <^7heo> the config has to be done in config.h 2014-11-07 12:21:44 <^7heo> and that's why you compile it yourself. 2014-11-07 12:22:02 <^7heo> and unless we have a binary format that allows editing the config directly in the binary, I WANT to keep it that way - no package. 2014-11-07 12:23:02 right, here i go, fresh install, setup-xorg-base, wish me luck 2014-11-07 12:23:27 ^7heo: you mean elf? 2014-11-07 12:24:13 <^7heo> or troll, I don't mind. 2014-11-07 12:24:13 adding entries (say, a new keybinding) would be a little tricky, everything else is fine 2014-11-07 12:24:19 <^7heo> ACTION overhides 2014-11-07 12:24:42 <^7heo> ovf: no, adding new entries wouldn't even be tricky. 2014-11-07 12:24:53 <^7heo> you just have to either store the size of your data section 2014-11-07 12:25:16 <^7heo> or to put a marker to end it. 2014-11-07 12:25:19 <^7heo> like EOF 2014-11-07 12:25:26 <^7heo> (since it can be stored after the binary) 2014-11-07 12:25:30 <^7heo> in ANY case 2014-11-07 12:25:47 <^7heo> you can totally compile a C code into a binary that has editable vars 2014-11-07 12:25:57 how do folks start x by hand - xinit or startx? echo windowmanager > .xinit ? 2014-11-07 12:26:08 <^7heo> and that way you could even be able to carry your programs + config when you statically compile, just by drag/dropping your stuff 2014-11-07 12:26:14 <^7heo> ScrumpyJack: xinit 2014-11-07 12:26:31 <^7heo> echo "dwm" >> .xinitrc 2014-11-07 12:26:36 <^7heo> and you're set. 2014-11-07 12:26:43 <^7heo> my login manager is: 2014-11-07 12:27:02 right, so setup-xorg-base + apk add xf86-video-intel twm should work right? 2014-11-07 12:27:20 <^7heo> yeah 2014-11-07 12:28:25 just moving to edge/testing first 2014-11-07 12:28:40 <^7heo> edge is the way to go 2014-11-07 12:28:42 <^7heo> with testing pinning 2014-11-07 12:29:29 arg, i forgot to pin 2014-11-07 12:29:48 oh well, we live dangerously 2014-11-07 12:30:20 upgrading install wayland 2014-11-07 12:31:05 <^7heo> I have to try wayland some day 2014-11-07 12:35:48 didnt work. i'll try edge only 2014-11-07 12:39:21 can you remind me the syntax for pinning in /etc/apk/repositories? 2014-11-07 12:39:49 <^7heo> @ http://... 2014-11-07 12:40:01 <^7heo> like, @testing http://... 2014-11-07 12:40:05 <^7heo> and then 2014-11-07 12:40:10 <^7heo> apk add stuff@testing 2014-11-07 12:41:13 ta 2014-11-07 12:41:20 <^7heo> ta? 2014-11-07 12:41:26 <^7heo> what language is that in? 2014-11-07 12:41:30 lol 2014-11-07 12:41:35 it means "thanks" 2014-11-07 12:41:38 <^7heo> ooh 2014-11-07 12:41:50 <^7heo> okay 2014-11-07 12:41:54 English slang basically 2014-11-07 12:42:30 <^7heo> I understand ty 2014-11-07 12:42:33 <^7heo> or thx 2014-11-07 12:42:55 <^7heo> but ta? 2014-11-07 12:43:02 <^7heo> ta ta maybe. but for something else. 2014-11-07 12:43:17 yeah lol, that means "bye" 2014-11-07 12:43:36 English is awesomly confusing if you think about it 2014-11-07 12:43:37 <^7heo> yeah 2014-11-07 12:43:52 so it's best not to think about it too much :) 2014-11-07 12:44:17 <^7heo> so, when you use english, it's best to not think too much. 2014-11-07 12:44:25 yes lol :) 2014-11-07 12:44:29 <^7heo> That explains a lot about the current situation of this planet. 2014-11-07 12:44:59 lol 2014-11-07 12:45:27 unfortunately that's also because the people in power who make decisions are seriously lacking in intelligence and/or common sense 2014-11-07 12:45:44 <^7heo> oh no they're not lacking any of that. 2014-11-07 12:45:58 <^7heo> they're just lacking interest in the people. 2014-11-07 12:46:20 <^7heo> rather, they are interested in more power, and sharing less. 2014-11-07 12:46:25 <^7heo> sharing is caring, that is true. 2014-11-07 12:47:26 indeed sharing is caring 2014-11-07 12:47:41 <^7heo> yeah well 2014-11-07 12:48:05 <^7heo> based on that, just look at the situation and you'll see. 2014-11-07 12:48:14 <^7heo> They're clever, and they make a lot of sense. 2014-11-07 12:48:19 <^7heo> They just are not caring. 2014-11-07 12:48:30 but what to do about it? 2014-11-07 12:48:39 <^7heo> use the power we have. 2014-11-07 12:48:45 <^7heo> ALL of it. 2014-11-07 12:48:47 <^7heo> brain 2014-11-07 12:48:49 <^7heo> time 2014-11-07 12:48:51 <^7heo> trust 2014-11-07 12:48:55 <^7heo> knowledge 2014-11-07 12:48:56 <^7heo> skill 2014-11-07 12:48:59 <^7heo> and use it well. 2014-11-07 12:49:09 <^7heo> build alternative solutions. 2014-11-07 12:49:13 <^7heo> and start using them. 2014-11-07 12:49:28 <^7heo> and build them with your brain, not with your hands. 2014-11-07 12:49:56 <^7heo> what I'm saying, here, to decrypt a bit is: 2014-11-07 12:50:05 <^7heo> money doesn't exist, paper and metal does. 2014-11-07 12:50:10 ok, we're in business. but my X resolution is crap 2014-11-07 12:50:15 <^7heo> so what is money, if it doesn't exist? 2014-11-07 12:50:22 suggestions on how to change it? 2014-11-07 12:50:25 <^7heo> Well, money is a physical representation of trust. 2014-11-07 12:50:35 Xorg -config or something? 2014-11-07 12:50:37 <^7heo> ScrumpyJack: xrandr 2014-11-07 12:50:48 <^7heo> and then 2014-11-07 12:50:53 <^7heo> we have internet 2014-11-07 12:50:55 <^7heo> it exists 2014-11-07 12:50:58 <^7heo> we know how it works 2014-11-07 12:51:02 ah, i think i just make that package :) 2014-11-07 12:51:06 <^7heo> we can now connect people, that are miles and miles away 2014-11-07 12:51:21 <^7heo> so in the end 2014-11-07 12:51:28 <^7heo> internet can provide better democracy 2014-11-07 12:51:39 any small X teminal to suggest? 2014-11-07 12:51:41 <^7heo> we can think before we spend money (aka trust) into something and someone. 2014-11-07 12:51:44 <^7heo> ScrumpyJack: st 2014-11-07 12:51:57 <^7heo> ScrumpyJack: git clone git://git.suckless.org/st 2014-11-07 12:52:08 in repos :) 2014-11-07 12:52:09 <^7heo> you'll need ncurses to make install 2014-11-07 12:52:14 <^7heo> in repos, nothing. 2014-11-07 12:53:45 hmm, lxterminal, gnome-terminal, xfce-terminal :( 2014-11-07 12:55:23 ^7heo: I shall continue to give this considerable thought, as always. Education of the populace is always a good idea methinks 2014-11-07 12:58:14 <^7heo> nice way to test webgl: http://www.vill.ee/eye/ 2014-11-07 12:58:40 <^7heo> ginjachris: ofc it is, but the best way to do it is to distribute. 2014-11-07 12:58:44 <^7heo> distributing is the future. 2014-11-07 12:58:48 <^7heo> and we need to do it right. 2014-11-07 12:58:56 <^7heo> distributing is sharing 2014-11-07 12:59:00 <^7heo> sharing is caring 2014-11-07 12:59:03 <^7heo> therefore distributing is caring. 2014-11-07 13:04:07 there is mrxvt 2014-11-07 13:04:27 xrandr is lying about my max res. i'll try later 2014-11-07 13:04:34 thanks for the help with X 2014-11-07 13:13:20 ScrumpyJack: if xrandr gets wrong data there's probably an issue this 'edid' data download. i've (SOMETIMES) had success if it was a screen connected via some analog port. put it on dvi, use nvidia tools to store the edid data, then put it back in its normal one, and force it to use the stored data 2014-11-07 13:13:30 oh and it took just like a day to make work 2014-11-07 13:13:37 maybe a day and a half 2014-11-07 14:00:00 <^7heo> I so love apk 2014-11-07 14:01:29 +1 love for apk 2014-11-07 14:02:34 I shall go so far as to say it is the finest package manager in the known universe 2014-11-07 14:02:47 (and probably the unknown too) 2014-11-07 14:04:00 <^7heo> it's incomplete 2014-11-07 14:04:06 <^7heo> but it's really great already 2014-11-07 14:04:53 <^7heo> brb, getting a club mate. 2014-11-07 14:08:00 <^7heo> and I really like busybox too. 2014-11-07 14:24:05 <^7heo> firefox customized with an array of plugins isn't SO bad. 2014-11-07 14:24:16 <^7heo> but you really need a LOT of those. 2014-11-07 14:25:32 <^7heo> like, adblock, ghostery, session manager, certificate patrol, open with, tree style tab, vimperator, and calomel SSL validation. 2014-11-07 14:25:35 <^7heo> At least. 2014-11-07 14:25:43 <^7heo> it's like... 8 plugins by DEFAULT. 2014-11-07 14:25:45 <^7heo> v_v 2014-11-07 14:31:23 yeah busybox is sweet 2014-11-07 14:31:32 ff is my browser of preference 2014-11-07 14:31:57 adblock, cookie monster, pentadactyl 2014-11-07 14:32:07 what's ghostery/cert patrol/calomel? 2014-11-07 14:32:21 oh, and noscript 2014-11-07 14:37:49 privacy badger :) https://www.eff.org/privacybadger 2014-11-07 14:39:21 ghostry is alright though, see https://www.ghostery.com/en/ 2014-11-07 14:40:29 https://addons.mozilla.org/en-US/firefox/addon/cipherfox/ <-- cipherfox is useful 2014-11-07 14:53:50 <^7heo> is pentadactyl better than vimperator? 2014-11-07 14:53:55 <^7heo> I remember it to be worse... 2014-11-07 14:54:44 <^7heo> yeah cipherfox seems better than calomel 2014-11-07 14:54:59 <^7heo> as calomel displays information in "color" which doesn't mean much, security wise. 2014-11-07 14:57:27 <^7heo> I wonder what the EFF means by "Several of these extensions have business models that we weren't entirely comfortable with." 2014-11-07 14:57:33 <^7heo> Since when extensions have business models? 2014-11-07 14:58:07 collecting data? ads with tracking? i dunno 2014-11-07 14:58:25 <^7heo> oh 2014-11-07 14:58:27 <^7heo> https://www.ghostery.com/en/how-we-make-money 2014-11-07 14:59:23 "Totally anonymous" "We hash out IP addresses" 2014-11-07 14:59:24 fail 2014-11-07 14:59:31 lol 2014-11-07 15:00:03 <^7heo> yeah 2014-11-07 15:00:11 <^7heo> Good thing I checked 2014-11-07 15:00:27 it's off by default tho, at least that's what they say 2014-11-07 15:00:33 but still, wtf 2014-11-07 15:00:39 just omit ip address completely 2014-11-07 15:01:07 <^7heo> yeah 2014-11-07 15:01:10 <^7heo> and country 2014-11-07 15:01:12 <^7heo> ffs 2014-11-07 15:01:24 <^7heo> I will analyze the network when I'm @home 2014-11-07 15:01:27 <^7heo> to check what they say 2014-11-07 15:01:55 <^7heo> but hashing an IP... 2014-11-07 15:01:56 <^7heo> wtf. 2014-11-07 15:02:11 <^7heo> are they aware that even with the best hash algo 2014-11-07 15:02:16 first of all the space is 32-bit 2014-11-07 15:02:25 <^7heo> calculating 2^32 hashes is VERY fast... 2014-11-07 15:02:25 so it can be brute forced in a few hours no matter how good the hash is 2014-11-07 15:02:43 <^7heo> I can calculate 2^32 hashes VERY quick here. 2014-11-07 15:02:45 <^7heo> I have like what 2014-11-07 15:02:50 <^7heo> 20 computers 2014-11-07 15:02:53 <^7heo> gigabit link 2014-11-07 15:03:01 <^7heo> and each computer has an i7 3.5GHz 2014-11-07 15:03:21 <^7heo> really, name your hash, and I give you the data in tar.bz2 :D 2014-11-07 15:03:23 ahem....I'm always open to donations. 2014-11-07 15:03:24 (lol) 2014-11-07 15:03:27 i have ssh to gccfarm 2014-11-07 15:03:35 <^7heo> same story then 2014-11-07 15:03:36 24-core amd64 boxes :) 2014-11-07 15:03:40 <^7heo> yeah 2014-11-07 15:03:45 <^7heo> so really 2014-11-07 15:03:47 <^7heo> hashing the IP 2014-11-07 15:03:52 <^7heo> sorry but that's a joke. 2014-11-07 15:04:11 even if it weren't easy to reverse, it's still an identifier 2014-11-07 15:04:16 <^7heo> hashing IPv6 would maybe make sense 2014-11-07 15:04:22 <^7heo> yeah 2014-11-07 15:04:24 <^7heo> even salted. 2014-11-07 15:04:34 you know "the same person who visited my-personal-site.com also visited hardcorexxx.com" 2014-11-07 15:05:04 you found me! 2014-11-07 15:05:10 scale that up to 1000 sites rather than 2 2014-11-07 15:05:17 <^7heo> yeah 2014-11-07 15:05:18 and you easily have an identifiable profile of who the person is 2014-11-07 15:05:20 <^7heo> and you have tracking 2014-11-07 15:05:27 <^7heo> right 2014-11-07 15:23:22 ^7heo: i'm not sure bz2 can be 4 bytes long 2014-11-07 15:23:52 ? 2014-11-07 15:25:07 ^7heo suggested he can unhash a .bz2 file in 2^32 hash calculations. this works out to 4 bytes. 2014-11-07 15:26:55 i think you misread, or he was joking 2014-11-07 15:27:00 not sure which since that statement was unclear 2014-11-07 15:27:14 but the topic was reversing hashes of ip (v4) addresses, which is a 32-bit space 2014-11-07 15:27:30 <^7heo> basically 2014-11-07 15:27:59 <^7heo> with 2^32 hashes of (say sha512) 512 bytes, you hardly get 4 bytes... 2014-11-07 15:50:58 <^7heo> 512 bit even 2014-11-07 15:52:06 <^7heo> it's going to be 256GB of uncompressed data. 2014-11-07 15:52:20 <^7heo> so if you can make that fit in 4 bytes 2014-11-07 15:52:23 <^7heo> I'm interested 2014-11-07 15:52:43 <^7heo> excepted if those four bytes are "null" 2014-11-07 16:04:16 ^7heo: pentadactyl used to be worse, now it's better, but I haven't used vimperator in like... 6mo+ 2014-11-07 16:13:24 <^7heo> on the other hand, it's not possible to use it with 33 2014-11-07 16:13:26 <^7heo> so... 2014-11-07 16:14:04 ^7heo: u can use the nightly version of pentadactyl 2014-11-07 16:14:09 that's what i do 2014-11-07 16:18:18 <^7heo> pnutzh4x0r: ah yeah 2014-11-07 16:18:21 <^7heo> pnutzh4x0r: does it work fine? 2014-11-07 16:21:06 it does on archlinux... 2014-11-07 16:21:11 :S 2014-11-07 16:21:37 im only running alpine on two servers, my laptop is still archlinux 2014-11-07 16:23:08 <^7heo> omg 2014-11-07 16:23:16 <^7heo> getoutofhere! 2014-11-07 16:23:20 <^7heo> raus! 2014-11-07 16:23:23 <^7heo> now! 2014-11-07 16:23:27 <^7heo> ACTION hides 2014-11-07 16:23:53 <^7heo> nah really, running arschlinux, I hesitate between "how can you" and "hats off for the nerves" 2014-11-07 16:26:06 hi 2014-11-07 16:26:31 <^7heo> moin 2014-11-07 16:27:30 if u keep a minimal configuration, archlinux isnt so bad 2014-11-07 16:27:40 it's a lot of little papercuts rather than big breaks 2014-11-07 16:27:49 <^7heo> yeah, if you don't touch the system, it's not so bad. 2014-11-07 16:27:56 <^7heo> You know what else isn't so bad if you don't touch it? 2014-11-07 16:27:58 <^7heo> drugs. 2014-11-07 16:28:01 <^7heo> ubuntu. 2014-11-07 16:28:02 <^7heo> windows 2014-11-07 16:28:06 <^7heo> MacOS X. 2014-11-07 16:28:16 <^7heo> ebola infected razorblades. 2014-11-07 16:28:23 <^7heo> and the list goes on 2014-11-07 16:28:33 im using ubuntu as the base for a programming contest environment... it's actually not that bad 2014-11-07 16:28:36 for that purpose anyways 2014-11-07 16:28:51 <^7heo> ubuntu works out of the box. 2014-11-07 16:28:54 i wouldnt run it on my personal machines 2014-11-07 16:28:54 <^7heo> and only that way. 2014-11-07 16:29:02 right, and that's y it's our base 2014-11-07 16:29:14 <^7heo> if you touch ANYTHING 2014-11-07 16:29:23 <^7heo> you have some non null chance to break it. 2014-11-07 16:29:30 <^7heo> same goes for arschlinux. 2014-11-07 16:29:43 <^7heo> the only difference between the two is that my mother can use noobuntu. 2014-11-07 16:30:11 <^7heo> arsch being for the script-hipster-kiddies 2014-11-07 16:30:25 yeah, but arch requires you to know something about your systerm 2014-11-07 16:30:31 so when it breaks, u can fix it 2014-11-07 16:30:35 or should be able to anyways 2014-11-07 16:30:42 <^7heo> well 2014-11-07 16:30:57 that's funny, b/c when i started using arch 8 years ago that's what my debian friend said 2014-11-07 16:30:59 <^7heo> arsch requires you to know something about UNIX so when it breaks, you know it's the fault of the shitty devs 2014-11-07 16:31:03 it's just the cool new hipster distro 2014-11-07 16:31:10 8 years later, people are still saying the same thing 2014-11-07 16:31:24 <^7heo> except that 8 years ago, arsch wasn't so bad. 2014-11-07 16:31:33 <^7heo> it was making sense 2014-11-07 16:31:37 i agree 2014-11-07 16:31:38 <^7heo> and was done with much more care. 2014-11-07 16:31:55 <^7heo> and I wasnt calling it arschlinux. 2014-11-07 16:31:56 arch stopped being a distro in the sense that they stopped integrating 2014-11-07 16:32:12 they now do w/e upstream tells them 2014-11-07 16:32:25 <^7heo> I expect the maintainers of a distro to know what they're doing 2014-11-07 16:32:31 <^7heo> and that rules out a LOT of linuxes for me. 2014-11-07 16:32:47 gtg, bye all, have a great weekend! 2014-11-07 16:32:50 <^7heo> no ubuntu, no mint, no debian, no arch, no ; just alpine. 2014-11-07 16:32:52 <^7heo> ginjachris: o/ 2014-11-07 16:32:57 cya 2014-11-07 16:33:13 <^7heo> debian is less broken than the rest on average... 2014-11-07 16:33:29 <^7heo> but it's bloated the same amount, and integrates as much shit. 2014-11-07 16:34:13 i think debian over does the integration 2014-11-07 16:34:25 but at least they try to be coherent 2014-11-07 16:34:28 <^7heo> yeah 2014-11-07 16:34:34 <^7heo> they turn it into total shit. 2014-11-07 16:34:45 <^7heo> that's a good way to be coherent without too much effort. 2014-11-07 16:35:17 how do u do networking on arch? networkd? dhcpcd? networkmanager? dhclient? netctl? 2014-11-07 16:35:37 and everything is in a constant state of deprecation 2014-11-07 16:36:10 <^7heo> yeah 2014-11-07 16:36:29 dont use netctl, use networkd... but networkd is new doesnt support all the features u need... oh well 2014-11-07 16:36:38 <^7heo> everything but systemd, full-bloated bash and as much gnu shit as possible to pack on a system that boots in less than 20 minutes. 2014-11-07 16:36:58 <^7heo> that never gets deprecated, because they need it. 2014-11-07 16:37:32 anyways, i plan on switching my laptop to alpine when my semester is over 2014-11-07 16:37:50 at least make an attempt anyways 2014-11-07 16:37:50 <^7heo> Good move. 2014-11-07 16:37:58 <^7heo> it's not like it's hard. 2014-11-07 16:38:04 <^7heo> All my machines run alpine. 2014-11-07 16:38:20 <^7heo> I just can't play on Linux yet. 2014-11-07 16:38:25 my laptop's running alpine 2014-11-07 16:38:28 <^7heo> 'cause steam isn't running on my alpine yet. 2014-11-07 16:38:43 well, i do have alpine in another partition, and some things are wonky 2014-11-07 16:39:06 like i have to use the vanilla kernel b/c grsec prevents i3status from reading stuff in /sys 2014-11-07 16:39:18 <^7heo> you're doing it wrong. 2014-11-07 16:39:21 unless i chmod it 2014-11-07 16:39:38 <^7heo> chmod the group 2014-11-07 16:39:42 <^7heo> be part of the group 2014-11-07 16:39:48 <^7heo> it's better than using vanilla. 2014-11-07 16:39:50 right, but that doesnt persist 2014-11-07 16:40:08 wtf 2014-11-07 16:40:09 Error relocating /usr/lib/libMagick++-6.Q16.so.5: KuwaharaImage: symbol not found 2014-11-07 16:40:22 looks like a missing library version dependency 2014-11-07 16:40:44 <^7heo> pnutzh4x0r: or you can suid your binary... but that's maybe not too great. 2014-11-07 16:41:18 i think i just need to write some scripts to chmod /sys on wakeup from suspend and bootup 2014-11-07 16:41:23 but there are other little papercuts 2014-11-07 16:41:30 <^7heo> dalias: that happens in edge yes. 2014-11-07 16:41:40 i also need to figure out a chroot/container so i can run chromium or flash 2014-11-07 16:41:51 ^7heo, do you know the cause? 2014-11-07 16:41:57 <^7heo> pnutzh4x0r: docker is for that. 2014-11-07 16:42:06 dalias: is this the latest edge package? 2014-11-07 16:42:19 yes 2014-11-07 16:42:39 <^7heo> dalias: seems like the dependency you pull is not in the required version 2014-11-07 16:43:31 <^7heo> just try to abuild the right version and install it 2014-11-07 16:43:33 <^7heo> to see if it helps. 2014-11-07 16:44:02 dalias: what did u run to get that error? 2014-11-07 16:45:42 i just installed imagemagic, and I don't have /usr/lib/libMagick++-6.Q16.so.5 2014-11-07 16:48:27 pnutzh4x0r, inkscape 2014-11-07 16:48:55 i don't even know what lib it's expecting to find that symbol in, tho 2014-11-07 16:49:22 perhaps magickcore or magickwand 2014-11-07 16:49:27 but which package are those in? 2014-11-07 16:50:08 maybe inkscape needs a rebuild? 2014-11-07 16:50:27 we just updated imagemagick yesterday 2014-11-07 16:50:32 <^7heo> not inkscape. 2014-11-07 16:50:35 <^7heo> okay 2014-11-07 16:50:39 <^7heo> dalias: please execute 2014-11-07 16:50:51 <^7heo> $ apk info --who-owns /usr/lib/libMagick++-6.Q16.so.5 2014-11-07 16:50:54 <^7heo> and then 2014-11-07 16:51:01 <^7heo> get that package APKBUILD from the git repo 2014-11-07 16:51:02 <^7heo> and then 2014-11-07 16:51:05 <^7heo> abuild -r it 2014-11-07 16:51:14 <^7heo> after having changed the version to the lastest one. 2014-11-07 16:51:22 <^7heo> and then apk add the package you just created. 2014-11-07 16:51:28 <^7heo> and see if it helps. 2014-11-07 16:52:48 ^7heo, that lib isn't the problem 2014-11-07 16:52:54 that lib needs a symbol from another lib 2014-11-07 16:52:56 and can't find it 2014-11-07 16:53:06 so the lib it wants to get that symbol from is the one that's outdated 2014-11-07 16:53:25 <^7heo> yeah but if you past half messages... 2014-11-07 16:53:49 ? 2014-11-07 16:53:56 that's the whole message 2014-11-07 16:53:58 <^7heo> it must mention what lib was required, right? 2014-11-07 16:54:07 no, ELF doesn't work that way 2014-11-07 16:54:16 <^7heo> it does it here, so... 2014-11-07 16:54:21 symbol references do not have an associated library they're expected to come from 2014-11-07 16:54:23 <^7heo> I don't know why you say that. 2014-11-07 16:54:26 the namespace is global 2014-11-07 16:54:27 <^7heo> I know that 2014-11-07 16:54:35 <^7heo> just nm to see that. 2014-11-07 16:54:48 i have a list of _all_ the libs libMagick++-6.Q16.so.5 depends on (DT_NEEDED) 2014-11-07 16:55:03 but no indication of which one KuwaharaImage is supposed to come from 2014-11-07 16:55:09 <^7heo> yeah, then grep U 2014-11-07 16:55:15 <^7heo> and you see the ones that are unresolved. 2014-11-07 16:55:28 *nod* 2014-11-07 16:55:33 KuwaharaImage 2014-11-07 16:55:34 <^7heo> but you already know which one is missing 2014-11-07 16:55:35 <^7heo> so... 2014-11-07 16:55:45 i would guess maybe one of these 2014-11-07 16:55:45 libMagickWand-6.Q16.so.2 => /usr/lib/libMagickWand-6.Q16.so.2 (0x7fc6043c2000) 2014-11-07 16:55:48 libMagickCore-6.Q16.so.2 => /usr/lib/libMagickCore-6.Q16.so.2 (0x7fc604041000) 2014-11-07 16:55:52 <^7heo> moment 2014-11-07 16:56:44 hm cant these dependencies be checked automatically at package build time? 2014-11-07 16:57:01 <^7heo> what if you ldd the so? 2014-11-07 16:57:12 <^7heo> nsz: normally they do, afaik 2014-11-07 16:57:43 or there could be a tool that does inverted index over all dsos 2014-11-07 16:57:54 <^7heo> dalias: I think it's in /usr/lib/libMagickCore-6.Q16.so.2 2014-11-07 16:57:58 <^7heo> you might want to rebuild that one. 2014-11-07 16:58:07 symbol -> dsolist with package information 2014-11-07 16:58:10 <^7heo> nsz: maybe the symbol is declared as weak 2014-11-07 16:58:20 <^7heo> and I dunno if abuild will see the dep in that case. 2014-11-07 16:59:43 nsz, the problem is it's not known what version the symbol was introduced in 2014-11-07 16:59:47 in this specific case.... 2014-11-07 16:59:56 both libs come from the same source package but they're different binary packages 2014-11-07 17:00:06 <^7heo> ah yeah 2014-11-07 17:00:08 <^7heo> makes sense 2014-11-07 17:00:11 hmm 2014-11-07 17:00:13 so the developers don't version it properly :( 2014-11-07 17:00:29 if i just run inkscape it works, at least it says nothing to do 2014-11-07 17:00:31 the imagemagick-c++ package should just always depend on exact-same-version imagemagick package 2014-11-07 17:00:49 <^7heo> that wouldn't happen with static compiling. 2014-11-07 17:00:53 since they treat the libs as internal from a versioning perspective 2014-11-07 17:01:01 strings /usr/lib/libMagickCore-6.Q16.so.2 | grep Kuwa returns KuwaharaImage 2014-11-07 17:01:02 i see 2014-11-07 17:01:47 pnutzh4x0r, mine doesn't. wtf 2014-11-07 17:02:07 apk info --who-owns /usr/lib/libMagickCore-6.Q16.so.2 2014-11-07 17:02:08 /usr/lib/libMagickCore-6.Q16.so.2 is owned by imagemagick-6.8.9.5-r0 2014-11-07 17:02:39 and there doesn't seem to be a newer version 2014-11-07 17:02:58 ah 2014-11-07 17:03:05 i have a newer version 2014-11-07 17:03:16 maybe the imagemagick-c++ didnt get updated? 2014-11-07 17:03:17 apk add -U imagemagick 2014-11-07 17:03:19 does nothing 2014-11-07 17:03:22 what version of imagemagick do u have 2014-11-07 17:03:28 see there 2014-11-07 17:03:30 6.8.9.5 2014-11-07 17:03:54 do u have edge enabled? 2014-11-07 17:03:56 the c++ package is 6.8.9.10 2014-11-07 17:03:58 yes 2014-11-07 17:04:27 so ur c++ version is the latest, but imagemagic itself is not 2014-11-07 17:04:30 weird 2014-11-07 17:05:09 yes 2014-11-07 17:05:22 maybe a mirror issue? 2014-11-07 17:05:23 http://mirrors.gigenet.com/alpinelinux/edge/main/x86_64/ 2014-11-07 17:05:27 i use that it has the latest 2014-11-07 17:05:35 my apk repos config is: 2014-11-07 17:05:35 @stable http://nl.alpinelinux.org/alpine/v3.0/main 2014-11-07 17:05:35 http://nl.alpinelinux.org/alpine/edge/main 2014-11-07 17:06:11 is imagemagick pinned at stable? 2014-11-07 17:06:18 ACTION has no idea how pinning works 2014-11-07 17:06:19 it shouldn't be 2014-11-07 17:06:35 i just pinned xorg because edge has a version with broken intel driver 2014-11-07 17:07:04 ok, so the package is there, but apk is not updating it 2014-11-07 17:47:48 clandmeter, uselessd-7 :3 2014-11-07 17:49:49 pnutzh4x0r, any idea how to diagnose that? 2014-11-07 20:31:23 dalias: hey 2014-11-07 20:32:07 yes? 2014-11-07 20:32:13 so I glanced at the nix source code 2014-11-07 20:32:22 annnnd there is like 0 design/specification documents 2014-11-07 20:32:32 :-p 2014-11-07 20:32:35 well none that I could find for now 2014-11-07 20:33:04 there are plenty of user manuals and articles 2014-11-07 20:33:21 so I think I am going to have to read through all the academic papers 2014-11-07 20:33:27 and then go from there 2014-11-07 20:39:35 if you really want to learn about Linux you should go through a gentoo build 2014-11-07 20:39:54 wooops wrong message box 2014-11-07 20:40:11 :P 2014-11-07 20:53:01 http://fun.irq.dk/funroll-loops.org/ 2014-11-07 20:53:05 :) 2014-11-07 20:56:25 LOL 2014-11-07 20:57:00 <3 gentoo 2014-11-07 20:58:17 50-50 for me 2014-11-08 11:13:56 in what package is the manual for all other packages 2014-11-08 11:14:07 man sudo No manual entry for sudo 2014-11-08 11:19:01 dash doc packages 2014-11-08 12:48:16 zxd: i have not tried this myself, but according to the wiki, you will get the man pages for package foo by running 'apk add foo-doc' 2014-11-08 12:48:37 accordingly then you will have to run 'apk add sudo-doc' 2014-11-08 12:49:03 but i usually do not install man-pages to save space on my flash 2014-11-08 17:26:10 hi huis 2014-11-08 17:26:15 hi guis :D 2014-11-08 17:26:37 *y 2014-11-08 17:26:41 exit 2014-11-08 17:46:39 hoi 2014-11-08 17:46:45 anyone using mcabber? 2014-11-08 18:52:06 ncopa: mcabber broken for me :( 2014-11-08 18:52:44 Disconnected, reason: 3->'LM_DISCONNECT_REASON_ERROR' 2014-11-08 18:53:02 something around ssl 2014-11-08 18:56:13 :/ 2014-11-08 18:56:37 have you investigated the cause? can it be that it's an ssl config problem? or a dns problem, or what ever? 2014-11-08 18:57:14 yes 2014-11-08 18:57:37 the same config working on the host machine 2014-11-08 18:57:55 and it works there, but not on alpine? 2014-11-08 18:58:00 yes 2014-11-08 18:58:14 and used to it was working on alpine as well 2014-11-08 18:58:49 i've applied some updates 2014-11-08 18:59:09 openssl is not a dependency in the apkbuild config 2014-11-08 18:59:40 my guess is openssl changed and mcabber has to rebuild 2014-11-08 19:46:53 yepp 2014-11-08 19:47:07 loudmouth-dev depends on openssl-dev 2014-11-08 19:47:16 and it hasn't been rebuilt 2014-11-08 19:47:21 loudmouth-dev-1.4.3-r1.apk 2014-May-13 13:56:53 2014-11-08 19:47:21 openssl-dev-1.0.1j-r0.apk 2014-Oct-16 06:42:21 2014-11-08 20:18:09 hmm 2014-11-08 20:20:56 hmmm. kool find. 2014-11-08 20:22:33 i've just updated to 3.1_alpha 2014-11-08 20:24:04 but still wrong :/ 2014-11-08 21:54:30 the end of strace full of these: 2014-11-08 21:54:31 clock_gettime(CLOCK_REALTIME, {xx, yy}) = 0 2014-11-08 21:54:32 read(4, 0xzz, 5) = -1 EAGAIN (Resource temporarily unavailable) 2014-11-08 21:54:37 maybe a futex issue? 2014-11-08 23:29:58 btw is there any hope we'll be able to boot and have working nfs server on 3.0? 2014-11-09 20:54:53 Can someone tell me why memory cgroups are disabled in the alpine kernel? I'd need them.. 2014-11-09 20:55:58 Rebuild? :) 2014-11-09 20:57:54 Yeah, sure... but should i send a patch, too? 2014-11-09 20:57:58 jomat: they should be enabled 2014-11-09 20:58:12 http://git.alpinelinux.org/cgit/aports/tree/main/linux-grsec/kernelconfig.x86_64?h=3.0-stable <- CONFIG_CGROUPS=y 2014-11-09 20:59:30 CONFIG_MEMCG is not set 2014-11-09 21:00:51 Yeah... grep CONFIG_MEMCG /boot/config-grsec 2014-11-09 21:01:31 jzono1: http://git.alpinelinux.org/cgit/aports/tree/main/linux-grsec/kernelconfig.x86_64?h=3.0-stable#n147 2014-11-09 21:02:42 sorry, sloppy searching 2014-11-09 21:02:48 np :-) 2014-11-09 21:03:01 just submit a feature request 2014-11-09 21:03:59 (and in the meantime just rebuild. its easy-ish to create a custom kernel package) 2014-11-09 21:04:01 I'd like to avoid that if there's a good reason against them :-) 2014-11-09 21:39:10 easy-ish as in undocumented pfft ;) 2014-11-10 06:38:39 anyone got a simple command line mail program they recomend? I just need it to send mail from a web-form script. 2014-11-10 06:38:55 hey, mail 2014-11-10 06:38:57 or sendmail 2014-11-10 06:39:37 yea "mail" isn't something I can apk add apparently 2014-11-10 06:40:11 oh, effectively 2014-11-10 06:40:19 I never tried 2014-11-10 06:40:43 i know that sendmail comes with postfix 2014-11-10 06:41:16 there is /usr/bin/sendmail 2014-11-10 06:41:35 on alpine, this is the implementation of busybox 2014-11-10 06:41:45 which might be plainfully functionnal 2014-11-10 07:33:57 morning 2014-11-10 07:34:07 darkfader: afaik nfs server should work with alpine v3.0 2014-11-10 07:34:18 some components does not, like idmap 2014-11-10 07:34:33 but basic nfs server should work 2014-11-10 07:36:59 NameBrand: you could use busybox sendmail unless you need smtp auth or full MX 2014-11-10 07:37:15 if you need smtp auth you could use ssmtp 2014-11-10 07:38:19 jomat: re CONFIG_MEMCG 2014-11-10 07:38:20 http://cateee.net/lkddb/web-lkddb/MEMCG.html 2014-11-10 07:38:50 Note that setting this option increases fixed memory overhead associated with each page of memory in the system. By this, 8(16)bytes/PAGE_SIZE on 32(64)bit system will be occupied by memory usage tracking struct at boot. Total amount of this is printed out at boot. 2014-11-10 07:38:51 Only enable when you're ok with these trade offs and really sure you need the memory resource controller. Even when you enable this, you can set "cgroup_disable=memory" at your boot option to disable memory resource controller and you can avoid overheads. (and lose benefits of memory resource controller) 2014-11-10 07:39:44 the reason was that was never enabled was: "Only enable when you're ok with these trade offs and really sure you need the memory resource controller." 2014-11-10 07:41:17 jomat: if you need that feature, can you please file an issue for it? 2014-11-10 07:41:37 also put some good reason why you think this should be enabled by default for everyone 2014-11-10 07:42:07 I am kinda ok with enabling it since you can disable it on boot prompt 2014-11-10 07:42:34 ok 2014-11-10 07:42:36 oh 2014-11-10 07:42:57 │ This config option also selects MM_OWNER config option, which │ 2014-11-10 07:42:57 │ could in turn add some fork/exit overhead. 2014-11-10 07:44:01 jomat: please file an issue about it and set target v3.1.0 2014-11-10 08:03:00 ncopa: yea, on the alpine system I don't see any errors or attempts to actually send the mail. Looks like it's the php mail function that's being used. Troubleshooting multiple systems so it's taking time. 2014-11-10 08:05:29 same code on bsd sends the message but doesn't return the user to the completion page, alpine it returns to the completion page but doesn't send the message 2014-11-10 08:07:56 :D 2014-11-10 08:08:20 NameBrand: iirc the phpinfo() function show how php sends email 2014-11-10 08:08:36 i think you can configure it in /etc/php/php.ini 2014-11-10 08:09:14 I like my static pages more 2014-11-10 08:09:25 if it fork/exec sendmail, then it will likely not work out of the box (unless you install and configure postfix) 2014-11-10 08:10:35 i suppose you have an external smtp host? 2014-11-10 08:10:37 http://php.net/manual/en/mail.configuration.php 2014-11-10 08:11:25 yea there's an external smtp host available. 2014-11-10 08:11:43 so the only thing you need to do is make busybox sendmail work 2014-11-10 08:12:19 i think all you need to do for that is set SMTPHOST= in /etc/profile 2014-11-10 08:12:52 alternatively: apk add ssmtp and edit /etc/ssmtp.conf 2014-11-10 08:13:23 ha 2014-11-10 08:13:33 i even wrote a setup-mta script 2014-11-10 08:13:38 that will set up ssmtp for you 2014-11-10 08:13:53 NameBrand: you can probably just run: setup-mta 2014-11-10 08:14:00 morning 2014-11-10 08:14:30 morning 2014-11-10 08:15:27 guess I'll uninstall the postfix I just installed... but run updates first 2014-11-10 08:16:01 makes for a fun 'dev' environment... primary server is freebsd/lighttpd, secondary is alpine/lighttpd. Configs and data get rsync'd 2014-11-10 08:18:05 ncopa: how is the build coming on? :) stuff still failing? 2014-11-10 08:18:12 ncopa: looks like ssmpt just needed to be setup for the alpine to start working, now to figure out wtf is up with the freebsd side 2014-11-10 08:19:01 ScrumpyJack: build-3-1-x86_64: files from v141022-505-g2b7dbbf uploaded 2014-11-10 08:19:04 \o/ 2014-11-10 08:19:19 seeme like clandmeter fixed the remainders this weeked 2014-11-10 08:19:42 ncopa: the resulting mail headers do look rather different... since the alpine goes direct to the smtp, while the freebsd uses the local MTA 2014-11-10 08:20:04 i'll try flush patches in my inbox, tag new mkinitfs release 2014-11-10 08:20:35 maybe look at kernel config CONFIG_MEMCG 2014-11-10 08:20:45 and then try have v3.1 rc1 out today 2014-11-10 08:21:16 and do some general package maintenance 2014-11-10 08:21:23 update as much as possible 2014-11-10 08:30:42 \o/ 2014-11-10 08:50:57 <^7heo> moin moin 2014-11-10 08:57:48 hot damn, figured it out. php.ini setting that needed to be changed. 2014-11-10 09:19:36 ncopa: i need this in my rc.local: http://hastebin.com/ujacofosaw.1c 2014-11-10 09:20:01 and some other guy on here had asked me, he had the same problem 2014-11-10 09:20:12 till then i thought it's just an error on my end 2014-11-10 09:20:30 exportfs will give function not implemented without that 2014-11-10 09:20:55 nfs is in /etc/modules, too. it just stopped working on boot after going to 3.0 2014-11-10 09:33:22 darkfader: did you file a bug about it? 2014-11-10 09:37:54 is anybody using zfs on alpine ? 2014-11-10 09:40:23 k0r10n: not I... I use zfs on my FreeBSD file server 2014-11-10 09:41:08 there is problem to mount pool via zfs command 2014-11-10 09:41:22 only when set mountpoint to legacy 2014-11-10 09:45:50 ncopa: no because there was this "nfs is broken in some release note", iirc there is a bug about it though. i'll dig later tonight 2014-11-10 09:45:53 today 2014-11-10 09:48:22 great. thanks 2014-11-10 09:50:52 <^7heo> ncopa: any progress on 3.1? 2014-11-10 09:59:40 ^7heo: yes 2014-11-10 10:00:34 seems like all packages builds now 2014-11-10 10:00:41 i am working on mkinitfs 2014-11-10 10:01:36 <^7heo> ncopa: mplayer still crashes in edge 2014-11-10 10:01:59 <^7heo> so, I dunno if you're packing a different version for 3.1 but if not, you might want to fix it first ;) 2014-11-10 10:02:09 please file a bug for it 2014-11-10 10:02:33 <^7heo> okay 2014-11-10 10:02:39 hum 2014-11-10 10:17:01 <^7heo> ncopa: #3502 2014-11-10 10:18:39 Goodbox, the mail/sendmail binaries are usually provided by your MTA 2014-11-10 10:19:02 Which range from very very minimal to exim :D 2014-11-10 10:20:06 You probably want a small one. 2014-11-10 10:20:29 <^7heo> exim... 2014-11-10 10:21:56 I usually just use postfix for consistency 2014-11-10 10:22:12 One MTA on all the systems 2014-11-10 11:01:11 <^7heo> Did I timeout? 2014-11-10 11:01:26 ^7heo: yes 2014-11-10 11:01:30 <^7heo> thanks 2014-11-10 11:01:34 <^7heo> I suspect a kiddie. 2014-11-10 11:01:40 <^7heo> my client stopped answering 2014-11-10 11:01:49 <^7heo> like irc flood. 2014-11-10 11:37:55 hello 2014-11-10 11:39:46 <^7heo> moin 2014-11-10 12:09:08 hi 2014-11-11 08:14:36 moring 2014-11-11 08:14:40 moring? 2014-11-11 08:14:42 morning 2014-11-11 08:15:15 moin 2014-11-11 08:19:52 moinmoin 2014-11-11 09:27:32 <^7heo> moin 2014-11-11 21:01:32 is anyone else having issues with nslookup - BusyBox v1.22.1 on Alpine 3.0.6 2014-11-11 21:01:51 it is also affects wget returning - bad address 2014-11-11 21:02:17 but when I use nslookup (coreutils) or dig on my laptop in the same network it works correctly 2014-11-11 23:21:56 also I am running a new lxc container and getting an IO after the APINDEX is fetched 2014-11-11 23:21:58 ERROR: http://nl.alpinelinux.org/alpine/v3.1/main/: IO ERROR 2014-11-11 23:22:19 I ran it again in verbose mode but got no extra informations 2014-11-11 23:22:30 Alpine LXC container from the stock template 2014-11-11 23:23:40 figured out the LXC issue 2014-11-12 08:39:32 morning 2014-11-12 10:02:30 heya 2014-11-12 10:03:28 i'm trying to use alpine linux as a base for my docker images, but have no luck getting apk to add packages inside it 2014-11-12 10:06:20 <^7heo> manveru: ? 2014-11-12 10:07:18 oh, nevermind, fixed it :) 2014-11-12 10:07:24 forgot to add apk update 2014-11-12 10:07:36 and just happened that all i tried to install was outdated 2014-11-12 10:08:41 finally got my image sizes under 20mb with alpine, really cool 2014-11-12 11:58:53 hi all 2014-11-12 12:01:49 @ncopa: you may wish to check I haven't broken bugs.alpinelinux.org - I was trying to be helpful and upload a bunch of valgrind output to Bug #3501 2014-11-12 12:02:00 I think it might have been a little too much for it 2014-11-12 12:02:06 My apologies :( 2014-11-12 12:02:23 2014-11-12 12:04:10 ncopa: the page finally loaded (http://bugs.alpinelinux.org/issues/3501) but you (or someone) might want to delete my last update......I'll add the output as a file instead. 2014-11-12 12:04:55 once again, sorry 2014-11-12 12:06:45 removed 2014-11-12 12:07:07 should i remove comment 2 too? 2014-11-12 12:07:10 thank you 2014-11-12 12:07:53 i didn't even realise there was more, so yes please 2014-11-12 12:08:02 I'll upload the outputs as a file 2014-11-12 12:08:09 might be a little more efficient :) 2014-11-12 12:10:16 and comment 1 too then i suppose 2014-11-12 12:10:29 done 2014-11-12 12:10:30 yeah, all the comments. I only see 1. 2014-11-12 12:10:38 thanks again. 2014-11-12 12:10:42 np 2014-11-12 12:10:45 before I touch anything.... 2014-11-12 12:10:57 ...is valgrind actually gonna help with this? 2014-11-12 12:11:04 maybe 2014-11-12 12:11:46 i can add a -dbg package for it 2014-11-12 12:11:57 so you can generate a coredump and analyze it 2014-11-12 12:13:31 well I can give it a go 2014-11-12 12:13:40 just be gentle with me lol 2014-11-12 12:14:01 np 2014-11-12 12:14:17 i wouldnt blame you for bugs.a.o either 2014-11-12 12:14:27 until today I had not used valgrind 2014-11-12 12:14:29 i think its bugs.a.o that should have rejected it 2014-11-12 12:14:35 valgrind is nice 2014-11-12 12:14:41 its great for tracking memleaks 2014-11-12 12:14:45 detecting 2014-11-12 12:14:45 the usage seems simlpe enough 2014-11-12 12:15:02 output not too bad, but the amount from siege has been considerable 2014-11-12 12:15:09 not a good one to start with! 2014-11-12 12:15:45 _stack_chk_fail () at src/env/_stack_chk_fail.c:18 2014-11-12 12:15:58 that normally means buffer overflow 2014-11-12 12:16:37 ^7heo found that one :) 2014-11-12 12:16:51 ok valgrind output is added 2014-11-12 12:16:59 [ ] siege-3.0.8.tar.gz 2014-09-05 15:11 485K 2014-11-12 12:17:08 maybe we upgrade before start troubleshooting more? 2014-11-12 12:17:12 btw it is not high priority if you have better things to do! 2014-11-12 12:17:18 sure thing 2014-11-12 12:17:26 gimme a few mins 2014-11-12 12:17:39 work gets in the way of interesting things :( 2014-11-12 12:21:18 i should have looked there first 2014-11-12 12:21:28 didn't realise package was on older version 2014-11-12 12:21:34 i am dumb sometimes 2014-11-12 12:30:04 grrr 2014-11-12 12:30:17 bunch of make errors 2014-11-12 12:30:41 and now customers want my help 2014-11-12 12:30:53 this could only be worse if I had to go do an exam tomorrow 2014-11-12 12:30:54 i pushed 3.0.8 2014-11-12 12:30:58 oh wait....I do! 2014-11-12 12:31:06 too late... 2014-11-12 12:31:10 You, Sir, are a magician! 2014-11-12 12:31:16 build error was due to u_int32_t 2014-11-12 12:31:28 change that to stdint type and it built 2014-11-12 12:31:37 see you are magic! 2014-11-12 12:34:25 Mister Copa, I shall attempt to give it a go later 2014-11-12 12:54:12 anyone remember where the IRC logs are? i need to collect some answers into an alpine wiki page 2014-11-12 12:58:33 got'em 2014-11-12 13:23:03 <^7heo> ginjachris: which one already? 2014-11-12 13:55:37 ScrumpyJack, is this what you mean:? 2014-11-12 13:55:39 http://irclogger.com/.alpine-devel/2014-11-12 2014-11-12 13:56:07 http://irclogger.com/.alpine-linux/2014-11-12 2014-11-12 14:03:47 i got them from here http://dev.alpinelinux.org/irclogs/ 2014-11-12 14:04:00 ^7heo: sorry I was on lunch....what have I broken now lol? 2014-11-12 14:05:18 alacerda: but thanks, that tool is very cool 2014-11-12 14:06:23 alacerda: so it's a bot that logs to irclogger? 2014-11-12 14:06:31 <^7heo> ginjachris: I dunno 2014-11-12 14:06:34 <^7heo> I need a pillow @work 2014-11-12 14:06:52 <^7heo> for now I can use my sweat shirt but I definitely need a pillo. 2014-11-12 14:06:55 <^7heo> pillow* 2014-11-12 14:07:44 good idea. Plus eyes painted on your eyelids. No one will ever know you are napping! 2014-11-12 14:11:31 <^7heo> ginjachris: nah I need a stand for my laptop to stay on my lap 2014-11-12 14:13:03 some sort of groin strap. Or maybe a custom codpiece? 2014-11-12 14:15:12 <^7heo> groin strap... 2014-11-12 14:15:21 <^7heo> please remove that image from my head 2014-11-12 14:15:27 lol 2014-11-12 14:16:56 codpiece it is then 2014-11-12 14:17:15 <^7heo> yeah, slightly better 2014-11-12 14:17:45 such a wonderful piece of attire 2014-11-12 14:23:43 won't that stand(whatever) block the air flow 2014-11-12 14:23:59 unless planning to get new laptop 2014-11-12 14:25:39 <^7heo> ?? 2014-11-12 14:27:53 ^7heo, if you use pillow/sweat-shirt as laptop stand 2014-11-12 14:28:12 <^7heo> vkrishn: you missed a part. 2014-11-12 14:28:57 ahh.. :) 2014-11-12 14:33:31 http://actor-framework.org/ interesting 2014-11-12 15:01:08 and call it libjublee, eeks 2014-11-12 15:01:16 maybe jublee-libs 2014-11-12 15:01:19 ;) 2014-11-12 15:03:24 darn, connection+xchat, useless 2014-11-12 15:03:36 recap, I think kdbus would have a huge impact 2014-11-12 15:03:53 was reading about bloom filtering and wondering if someone could hybrid libjudy+bloom logic 2014-11-12 15:04:08 and call it libjublee, eeks 2014-11-12 15:04:14 maybe jublee-libs 2014-11-12 15:04:17 ;) 2014-11-12 15:06:05 what's wrong with irc protocol, just cannot check if connection is ESTABLISHED before atleast posting 2014-11-12 15:10:21 ScrumpyJack, hehe good question! i don't know... 2014-11-12 15:10:26 :) 2014-11-12 15:15:04 ScrumpyJack, I had requested the log service sometime back 2014-11-12 15:15:41 kinda, very helpful with unstable 3G connection 2014-11-12 17:09:25 bye all, have a good evening! 2014-11-12 23:15:21 *cough* off topic... http://muchweb.me/systemd-nsa-attempt/ 2014-11-13 08:27:32 jomat, conspiracy theories aside, systemd _IS_ a huge attack surface 2014-11-13 08:27:47 That point is actually very much valid 2014-11-13 08:38:12 fr0stb1te: uselessd updated 2014-11-13 08:38:56 clandmeter, sweet, thx. 2014-11-13 09:11:32 <^7heo> fr0stb1te: what is it that you call an attack surface? 2014-11-13 09:16:16 ^7heo, seriously. http://en.wikipedia.org/wiki/Attack_surface 2014-11-13 09:17:47 <^7heo> I knew attack vectors 2014-11-13 09:18:03 <^7heo> but I was unaware that a set of attack vectors is called an attack surface. 2014-11-13 09:18:49 Ah. I thought you were being condescending and not actually asking. 2014-11-13 09:18:57 The Internet has corrupted me 2014-11-13 09:19:34 <^7heo> nah, arschlinux. 2014-11-13 09:19:57 <^7heo> it's hard to see sincerity when you're used to arrogants pricks behaving like condescendent trolls. 2014-11-13 09:21:00 I'm pretty sure for me it was 4chan. 2014-11-13 09:26:34 <^7heo> 4chan is much more mature and mentally sane than the arschlinux channel 2014-11-13 09:29:31 Depends on the board 2014-11-13 09:29:50 Thematic boards — sure, most are sane-ish 2014-11-13 09:30:02 With exceptions like /v/idia 2014-11-13 09:30:08 /v/idiya* 2014-11-13 09:30:21 /pol/ is a shitfest too pretty much 2014-11-13 09:31:25 <^7heo> /b/ is much better than #arschlinux anyway. 2014-11-13 09:31:49 <^7heo> ben when they post images of girls eating the poo of other girls and vomiting it into their private parts. 2014-11-13 09:31:57 <^7heo> s/ben/even. 2014-11-13 09:32:30 <^7heo> #arschlinux beeing the biggest pool of retards that in the whole known universe. 2014-11-13 09:34:18 http://arschlinux.de/ D: 2014-11-13 09:34:35 I've joined the channel to see if I could register it for mocking purposes 2014-11-13 09:35:00 Or. Actually. That could be the best name for a channel for my own arch spinoff 2014-11-13 09:35:09 But it already exists. 2014-11-13 09:39:10 <^7heo> haha 2014-11-13 09:39:32 I keep thinking that Spark needs a channel 2014-11-13 09:39:46 But it will probably be an empty channel. 2014-11-13 09:49:43 <^7heo> you know what needs a channel? 2014-11-13 09:49:53 <^7heo> /dev/null 2014-11-13 09:50:25 <^7heo> and I would like that everything that you type into /dev/null gets sent to the real /dev/null of the server. 2014-11-13 09:51:01 So, a channel you can't speak in? 2014-11-13 09:51:14 <^7heo> exactly. 2014-11-13 09:51:17 <^7heo> https://scontent-b-ams.xx.fbcdn.net/hphotos-xfp1/v/t1.0-9/10446498_10152803503536112_2962530603189116284_n.png?oh=b0b22447cd9653d0d76ddd151e12d776&oe=5518EEE3 2014-11-13 09:57:40 ^7heo, btw I've looked at suckless.org and made my blog actually look a bit more like their site. I like their header more than my old one. 2014-11-13 09:59:46 And I actually remembered where I snached the colours from. dwm. 2014-11-13 10:00:55 <^7heo> haha 2014-11-13 10:00:59 <^7heo> I told ya ;D 2014-11-13 10:04:42 Oh well. Spark has a channel now. Why not. 2014-11-13 10:05:26 <^7heo> irc channels are cheap. 2014-11-13 10:05:47 Well. It takes up a frame in my irssi 2014-11-13 10:06:19 <^7heo> I'm talking about the cost server-side. 2014-11-13 10:06:33 <^7heo> what you're doing with your client doesn't really concern them. 2014-11-13 10:06:36 Yeah, that's true. 2014-11-13 10:06:42 <^7heo> unless you're flooding/spamming. 2014-11-13 10:07:07 Well. Even so freenode has some pretty restrictive rules for channel creation 2014-11-13 10:07:52 <^7heo> I dunno 2014-11-13 10:07:59 <^7heo> I'd not make a channel here. 2014-11-13 10:08:06 I didn't :) 2014-11-13 10:08:10 It's on oftc 2014-11-13 10:09:00 <^7heo> good :) 2014-11-13 10:09:04 Also someone has grabbed my main nick here, so... 2014-11-13 10:09:29 <^7heo> what's your main nick? 2014-11-13 10:09:31 fbt. 2014-11-13 10:09:48 <^7heo> /dev/fbt 2014-11-13 10:10:10 librefbtd-ng :) 2014-11-13 10:10:20 systemd-fbtdctl! 2014-11-13 10:39:55 morning 2014-11-13 11:04:41 ive managed to stay away from systemd. i've never been interested. like NetworkManager in a way, only more so. Perhaps had systemd had been limited to the functionality of, say, SMF in solaris, then fine. but i still prefer init and 2014-11-13 11:06:46 I actually liked the ideas when it first was introduced in Arch 2014-11-13 11:06:55 alpine linux is like a warm blanket of comfort. i know exactly what's going on. i can see everything in detail and that gives me a warm fuzzy confidence 2014-11-13 11:07:25 But then I looked at it closely and went “whaaa-? Why does everything have to be in the project? Why would you...” 2014-11-13 11:07:31 until you update without reading changes and your kernel breaks 2014-11-13 11:07:32 And then it went downhill 2014-11-13 11:07:53 So now I have mo own Arch spinoff without systemd :D 2014-11-13 11:08:24 In a way, systemd forced me to learn more about linux systems, so that's good, I guess. 2014-11-13 11:08:24 I use ubuntu on my laptop with luks drive encryption 2014-11-13 11:12:59 Diftraku: I agree. i don't care what runs on the family laptops as long as its pretty and it keeps my girls from complaining that is doesn't work (currently elementaryOS) . everything else is moving to alpine linux and gentoo. 2014-11-13 11:30:11 ubuntu has that certain level of useability that windows does 2014-11-13 11:31:08 Right up to the point where something breaks and the user has no “friendly” ways of telling what's the problem 2014-11-13 11:31:41 The underlying problem of making comuters “friendly” is that you build up complexity very fast that way 2014-11-13 11:31:58 see gconf :3 2014-11-13 11:32:05 And “friendly” systems have so much more points of breakage. 2014-11-13 11:32:29 Which, at the actual moment where something breaks, leaves a “user” completely helpless 2014-11-13 11:33:14 And it's all an evil cycle of “make this more friendly — build up complexity — it broke — fix it with more complexity — repeat” 2014-11-13 11:33:24 It's just not worth it 2014-11-13 11:33:45 Would be much more productive to teach people about the systems they are using 2014-11-13 11:33:52 a foolproof "dumb user" os would run under a hypervisor with the gpu and peripherials connected to the guest 2014-11-13 11:34:07 upgrades would just load a new image and user only needs to restart 2014-11-13 11:34:24 if it breaks, user just restarta and fresh working snapshot is used 2014-11-13 11:34:42 daya is stored on a separate disk with a backup scheme in place 2014-11-13 11:35:14 ie. not raid 2014-11-13 11:40:32 There are still many ways that could break. An update could break everything. It can still break in runtime when the user changes something. It can be hacked 2014-11-13 11:40:50 Comuters are complex machines with complex software running on them. A lot of breakage points. 2014-11-13 11:41:10 Making software “friendly” on top of that is just ridiculous 2014-11-13 11:41:38 And yet the industry was trying to do so since the 90s now 2014-11-13 11:42:17 But the worst part is that they are trying to make inherently hacker-centric systems to be user-friendly now 2014-11-13 11:42:20 See: systemd 2014-11-13 11:42:32 And it's just bloody infuriating 2014-11-13 11:44:44 <^7heo> I find it tiring. 2014-11-13 11:44:51 <^7heo> but maybe it's because I need more sleep. 2014-11-13 11:57:27 One part of me is greatful to systemd. It has *considerably* narrowed my choice of linux distros in a crowded marketplace. I've gone from too much choice to just enough choice, and that makes me happy. 2014-11-13 11:58:44 That's one way to look at ot 2014-11-13 11:58:45 it* 2014-11-13 11:59:06 I like the controversy around it for jolting the community into doing werid things 2014-11-13 11:59:11 Weird things are cool 2014-11-13 11:59:16 <^7heo> ScrumpyJack: right :P 2014-11-13 12:00:51 Then there's the fact that systemd, in its own twisted way, does address some problems that major distros had forever now. 2014-11-13 12:01:15 So it made people think about how to properly address those. 2014-11-13 12:01:19 And that's also nice. 2014-11-13 12:01:58 <^7heo> yeah 2014-11-13 12:02:09 <^7heo> when nothing exists for too long 2014-11-13 12:02:15 <^7heo> something will inevitably happen 2014-11-13 12:02:20 <^7heo> and if that is wrong 2014-11-13 12:02:25 <^7heo> it will be replaced. 2014-11-13 12:02:28 <^7heo> look at windows... 2014-11-13 12:02:34 "doing werid things"? it's jolted me into believing further in a shell, vi, sed and awk 2014-11-13 12:02:37 Like. Yeah, sure, debian, rhel and some others had HORRIBLE init scripts. People who are in denial about that are just stupid. That's a problem. 2014-11-13 12:03:47 But is introducing non-deterministic dependency-based boot process with declarative configuration a good dolution? I don't think so either 2014-11-13 12:04:09 It's the classic problem of over-engineering a solution. 2014-11-13 12:04:51 ScrumpyJack, people are doing weird things with shell too. 2014-11-13 12:04:55 Me, for example :D 2014-11-13 12:07:21 re: "HORRIBLE init scripts" init worked fine without distros trying to introduce dependancies. "Start things in this order" was never broken for me. 2014-11-13 12:08:33 I've been an admin for some time now, those scripts work when you are not trying to do some weird thing with the service 2014-11-13 12:08:49 And I personally thing that dependencies are cool but should be used sparingly 2014-11-13 12:09:10 Just like service auto-respawning is a corner-case solution if ever. 2014-11-13 12:09:46 But all that doesn't invalidate the fact that traditional sysv init scripts in major distros were always horrible messes 2014-11-13 12:09:56 They are a pain to modify and debug 2014-11-13 12:10:11 Arch's initscripts were actually pretty nice btw. 2014-11-13 12:10:35 I'm not hating on the concept of init scripts in general. Hell, my own service manager is basically an init script framework 2014-11-13 12:11:16 And yeah, deterministic bood is really good for servers 2014-11-13 12:11:20 boot* 2014-11-13 12:11:48 Start things that need to start after other things in a specific order, start everything else in parallel in any fucking order — that's great 2014-11-13 12:12:08 That's transparent and easy to control 2014-11-13 12:12:16 [15:03:47] < openfbtd> But is introducing non-deterministic dependency-based boot process with declarative configuration a good dolution? I don't think so either 2014-11-13 12:12:28 But the scripts in debian are still fucking horrific 2014-11-13 12:14:22 RHEL too btw. 2014-11-13 12:14:33 And, by extension — CentOS 2014-11-13 12:16:33 i lost interest when insserv appeared 2014-11-13 14:03:49 so, i now have alpine linux running X on my work desktop! yay! 2014-11-13 14:04:21 what's a good editor with minimal dependancies? 2014-11-13 14:05:13 abiword is missing lots of buttons 2014-11-13 14:05:16 <^7heo> #define editor 2014-11-13 14:05:27 <^7heo> oh, a text processor... 2014-11-13 14:05:30 <^7heo> google docs. 2014-11-13 14:06:24 i need a browser for that :) 2014-11-13 14:06:27 <^7heo> yes. 2014-11-13 14:06:34 <^7heo> apk add firefox@testing 2014-11-13 14:06:40 <^7heo> 33.0.2 2014-11-13 14:06:40 i'm good 2014-11-13 14:06:52 <^7heo> otherwise, vi. 2014-11-13 14:07:01 i don't really need a browser on this yes 2014-11-13 14:07:05 <^7heo> but I guess it's going to be "missing a lot of buttons" 2014-11-13 14:07:10 s/yes/yet 2014-11-13 14:07:54 do you have abiword on an alpine box? 2014-11-13 14:09:22 ^7heo what do you think i mean by "missing a lot of buttons"? 2014-11-13 14:09:36 'not microsoft word' 2014-11-13 14:09:53 <^7heo> yeah, something along that line. 2014-11-13 14:10:23 <^7heo> and abiword is the biggest pile of scheiße in the known universe. 2014-11-13 14:10:32 <^7heo> and maybe also outside of it. 2014-11-13 14:11:04 systemd's bundling of udev is unfortunate 2014-11-13 14:11:18 because newer usbutils require a newer udev than we provide 2014-11-13 14:11:34 and i strongly suspect that udev-only is not a supported configuration of systemd ;) 2014-11-13 14:12:42 lol, no, that's not what i mean. i find that offensive actually. no, it's the toolkit playing up. i'll take a screenshot 2014-11-13 14:12:42 kaniini, eudev :) 2014-11-13 14:13:42 openfbtd: we are presumably considering that as an option, but it has problems 2014-11-13 14:14:29 Hmm. Such as? I'm curious. 2014-11-13 14:14:43 I'm personally using eudev on my machines in my arch spinoff. 2014-11-13 14:14:56 Well. On one of them. smdev is fine for others 2014-11-13 14:15:20 http://st.ilet.to/abi.png 2014-11-13 14:16:00 abiword in alpine 2014-11-13 14:16:30 We expect that clients will soonishly just start doing normal bus calls to the new udev, like they'd do them to any other system service instead of using libudev. 2014-11-13 14:16:38 going to eudev does not fix anything 2014-11-13 14:16:46 oh wow, that is f'd 2014-11-13 14:17:16 You might want to talk about that to the eudev team 2014-11-13 14:17:31 They may have thoughts on the situation 2014-11-13 14:17:42 #gentoo-eudev here. 2014-11-13 14:18:21 so, now i'm looking for a non-gtk editor :) 2014-11-13 14:20:13 <^7heo> kaniini: are you saying that systemd is becoming a hard dep of Linux itself? 2014-11-13 14:20:30 No, only of apps that want to use udev 2014-11-13 14:20:30 yes, it appears so. 2014-11-13 14:20:43 <^7heo> openfbtd: udev is now part of Linux 2014-11-13 14:20:48 <^7heo> afaik at least 2014-11-13 14:20:48 And it's systemd's mission statement, why are you surprised? 2014-11-13 14:20:57 <^7heo> https://en.wikipedia.org/wiki/Udev 2014-11-13 14:21:06 Eh. I run systems without udev 2014-11-13 14:21:13 <^7heo> yeah, on BSD. 2014-11-13 14:21:18 No, linux. 2014-11-13 14:21:20 <^7heo> Ah. 2014-11-13 14:21:23 it is possible to run without udev on linux 2014-11-13 14:21:29 <^7heo> Then let's do it. 2014-11-13 14:21:36 <^7heo> I'm sure I'll like it. 2014-11-13 14:21:38 Thing is, not that many apps actually use udev directly 2014-11-13 14:21:40 it is not possible to provide desktop linux without udev 2014-11-13 14:21:55 <^7heo> kaniini: why? 2014-11-13 14:22:02 those FooKit things depend on it 2014-11-13 14:22:12 <^7heo> wait 2014-11-13 14:22:21 DeviceKit, etc 2014-11-13 14:22:26 <^7heo> yeah but... 2014-11-13 14:22:31 kaniini, it is not possible to provide a “full DE experience” on linux without udev then 2014-11-13 14:22:33 which is what desktop environments use 2014-11-13 14:22:38 I use my desktops without udev 2014-11-13 14:22:48 <^7heo> the only thing I had that wanted to interact with udev, so far, was mesa. 2014-11-13 14:22:58 openfbtd: that is great that it works for you, but this isn't "openfbtd's linux distribution" 2014-11-13 14:23:01 mesa does not use udev 2014-11-13 14:23:01 And I'm not saying “let's not care about udev” 2014-11-13 14:23:02 <^7heo> openfbtd: same here I suspect, but I can't check. 2014-11-13 14:23:08 <^7heo> kaniini: the demos 2014-11-13 14:23:12 nope 2014-11-13 14:23:15 <^7heo> strange 2014-11-13 14:23:17 <^7heo> wait 2014-11-13 14:23:23 I'm just saying that udev isn't _that_ essetial _yet_. 2014-11-13 14:23:34 But yeah, as I've said, the eudev team might have ideas 2014-11-13 14:23:36 <^7heo> I don't have glxgears here 2014-11-13 14:23:36 Or plans 2014-11-13 14:23:43 <^7heo> yeah 2014-11-13 14:23:49 <^7heo> I wish I knew more about udev 2014-11-13 14:23:54 openfbtd: their only plan would be to implement a kdbus thing 2014-11-13 14:24:02 And they may do that. 2014-11-13 14:24:04 i see eudev as dead on arrival at this point 2014-11-13 14:24:16 I've poked blueness about this 2014-11-13 14:24:22 <^7heo> then what's the alternative to systemd? 2014-11-13 14:24:34 There is no alternative to systemd 2014-11-13 14:24:37 the systemd mission statement is to ensure that there is none 2014-11-13 14:24:55 It's an intentional moving target with tentacles in every major subsystem 2014-11-13 14:25:16 my point is that it is looking increasingly likely that alpine will at some point in time, be forced to adopt systemd 2014-11-13 14:25:34 not because we want to, but because otherwise we will have a exponentially higher maintenance burden 2014-11-13 14:25:35 Unless you state that you don't care about desktops at all. 2014-11-13 14:25:41 <^7heo> what's the systemd mission statement? 2014-11-13 14:25:56 <^7heo> okay 2014-11-13 14:25:59 ^7heo: to be *the* middleware between kernelspace and userspace 2014-11-13 14:26:03 <^7heo> then I'll really have to go for my own linux. 2014-11-13 14:26:09 ^7heo, in Lennart's own words, systemd “is THE plumbing project” 2014-11-13 14:26:09 <^7heo> ah 2014-11-13 14:26:21 As in, the one and only solution 2014-11-13 14:26:21 <^7heo> fuck 2014-11-13 14:26:28 <^7heo> that's harmful. 2014-11-13 14:26:31 openfbtd: what will we do when X requires kdbus 2014-11-13 14:26:46 <^7heo> let's replace trillions of things by one BIG set of interdependent binaries. 2014-11-13 14:26:47 that is coming eventually 2014-11-13 14:27:15 <^7heo> we need to change the approach here. 2014-11-13 14:27:29 <^7heo> distributions have to go P2P 2014-11-13 14:28:28 is udev required because existing apps/libs depend on it or is there some more fundamental reason? 2014-11-13 14:28:35 <^7heo> so then the "burden of maintenance" can be shifted away to the mass. 2014-11-13 14:28:50 nsz, existing apps/libs query libused for various things 2014-11-13 14:28:57 libudev* 2014-11-13 14:29:01 nsz: usbutils requires it 2014-11-13 14:29:50 at any rate, i would be very skeptical on the correctness of eudev's kdbus implementation, were it to gain one 2014-11-13 14:30:44 <^7heo> moin clandmeter 2014-11-13 14:33:16 kaniini, ah, I've also completely forgotten about evdev >_> 2014-11-13 14:33:20 Yeah, that's a problem 2014-11-13 14:34:23 <^7heo> what's usbutils for already? 2014-11-13 14:34:32 handling usb devics 2014-11-13 14:35:30 but if it's using libudev to communicate with the udev daemon then only libudev needs to be replaced 2014-11-13 14:35:47 nsz: lennart's vision is to remove libudev 2014-11-13 14:35:52 ah 2014-11-13 14:35:53 nsz: and use kdbus calls instead 2014-11-13 14:36:38 Also, why would eudev's implementation not correct? 2014-11-13 14:36:45 be not correct* 2014-11-13 14:37:27 ok so usbutils needs an alternative api to communicate with a daemon that can handle the uevents from the kernel 2014-11-13 14:37:44 nsz: they are planning on removing uevents too 2014-11-13 14:37:59 nsz: everything is planned to be going kdbus is the problem 2014-11-13 14:38:05 ah 2014-11-13 14:38:35 so it will be lennart-controlled all the way down 2014-11-13 14:38:36 openfbtd: because there's a lot of policy involved in setting up dbus 2014-11-13 14:38:47 openfbtd: selinux-type things etc too 2014-11-13 14:39:04 dbus is an overengineered pile of junk 2014-11-13 14:39:10 ACTION sighs 2014-11-13 14:39:19 Well. eudev basically follows upstream right now 2014-11-13 14:39:29 at this point, i am more interested in just porting alpine and musl to run on freebsd's kernelspace 2014-11-13 14:39:34 There is no reason they can't fetch that code from systemd directly 2014-11-13 14:39:51 it seemed to me there are serious issues with kdbus last time it hit the news.. 2014-11-13 14:40:58 (useless hierarchical namespace that requires special permissions to manage, no way to drop credentials and still keep a kdbus connection going) 2014-11-13 14:41:32 Yeah, it was basically rejected from the kernel for now 2014-11-13 14:41:35 it's ok, greg k-h is on the case to ensure it is forced through 2014-11-13 14:41:47 But the systemd crowd will just force it 2014-11-13 14:41:50 Yep 2014-11-13 14:44:04 Well 2014-11-13 14:44:10 Unless linux steps in 2014-11-13 14:44:14 Linus* 2014-11-13 14:44:27 That's basically the only hope. 2014-11-13 14:44:37 linus does not care 2014-11-13 14:44:40 he likes systemd 2014-11-13 14:44:52 He is indifferent to systemd 2014-11-13 14:45:01 And he also hates bad code and bad solutions 2014-11-13 14:45:19 Which might result in him saying “kdbus is horrible, make it properly or fuck off” 2014-11-13 14:45:25 Or it might not. 2014-11-13 14:45:33 it wont 2014-11-13 14:45:39 redhat will just pay him to shut up 2014-11-13 14:46:11 I'm pretty sure that redhat doesn't have that much influence over him 2014-11-13 14:46:25 you are way too optimistic 2014-11-13 14:53:54 <^7heo> yeah 2014-11-13 14:54:01 <^7heo> people like money 2014-11-13 14:54:06 <^7heo> especially rich people 2014-11-13 14:54:19 <^7heo> because they already dream so, it's not hard to make them dream more. 2014-11-13 14:59:28 <^7heo> I just want to die. 2014-11-13 15:00:00 hi ^7heo 2014-11-13 15:00:02 dispair or depression 2014-11-13 15:00:08 and the rest 2014-11-13 15:00:17 hi dude 2014-11-13 15:00:39 <^7heo> ScrumpyJack: systemd. 2014-11-13 15:02:25 bring back hotplug 2014-11-13 15:02:39 hm is it true that the udevd<->kernel and udevd<->application apis are both planned to changed to kdbus? 2014-11-13 15:04:10 this is what lennart said yes 2014-11-13 15:04:26 http://lists.freedesktop.org/archives/systemd-devel/2014-May/019657.html 2014-11-13 15:05:21 then how is it udev anymore if all external apis are changed? :) 2014-11-13 15:05:37 (other than the configuration mess presumably) 2014-11-13 15:08:45 <^7heo> so the whole point of systemd is "hiding technical hacks behind an opaque hood" 2014-11-13 15:08:59 <^7heo> and people like it because they can imagine that the system is "cleaner" 2014-11-13 15:17:09 nsz, that's basically it 2014-11-13 15:17:14 <@kaniini> at this point, i am more interested in just porting alpine and musl to run on freebsd's kernelspace 2014-11-13 15:17:39 dalias: yes, the state of linux is seriously depressing 2014-11-13 15:17:42 kaniini, is there a reason you don't want to do this via the linux syscall api and just minor tweaks for kernel things that aren't part of the syscall api? 2014-11-13 15:18:10 because freebsd offers linux syscall api and it's much better than the freebsd one :) 2014-11-13 15:18:13 dalias: i think it is possible to get higher performance by using freebsd primitives directly, for things like threads 2014-11-13 15:18:45 it may be, but the question is whether it's possible to get correct behavior 2014-11-13 15:19:01 i'm not optimistic when it comes to bsd's and threads :( 2014-11-13 15:19:25 if you have any docs i could look at on the low-level freebsd primitives, i could try to make an assessment of that 2014-11-13 15:20:20 <^7heo> if the world is going SO wrong... 2014-11-13 15:20:26 <^7heo> it's me who must be wrong. 2014-11-13 15:21:26 <^7heo> as Descartes said: "Rather change your desires than the world order." 2014-11-13 15:21:28 anyway re: kdbus invading udevd, udevd is already junk 2014-11-13 15:21:31 it was consumed by systemd 2014-11-13 15:21:43 there's eudev and busybox mdev that replace it 2014-11-13 15:21:49 <^7heo> I don't want to do anything with computers anymore. 2014-11-13 15:21:55 <^7heo> I'm going to become a ranger 2014-11-13 15:21:56 <^7heo> or something 2014-11-13 15:22:10 <^7heo> wish me luck. 2014-11-13 15:22:26 so i don't think it's much of a big deal what systemd does there with their junk udev implementation 2014-11-13 15:22:39 the bigger issue is user-facing apps that want to talk with systemd using this crap 2014-11-13 15:22:47 <^7heo> exactly. 2014-11-13 15:22:54 <^7heo> the more apps switch to systemd 2014-11-13 15:22:57 <^7heo> the more we're fucked. 2014-11-13 15:23:03 <^7heo> because we will have a kernel 2014-11-13 15:23:08 <^7heo> and nothing working on top of it. 2014-11-13 15:23:14 <^7heo> and the kernel will have systemd implemented in it anyway 2014-11-13 15:23:18 nothing? or no useless gnome shit? 2014-11-13 15:23:24 <^7heo> nothing. 2014-11-13 15:23:27 <^7heo> I mean 2014-11-13 15:23:27 huh? 2014-11-13 15:23:34 <^7heo> as soon as linux contains systemd 2014-11-13 15:23:43 <^7heo> they will be linked together in source 2014-11-13 15:23:44 the vast majority of apps have absolutely no reason, not even any made-up stupid fdo.org reason, to communicate with systemd 2014-11-13 15:24:05 tinfoil hatting does not do your cause any service 2014-11-13 15:24:31 <^7heo> didn't get anything after tinfoil hatting 2014-11-13 15:24:41 <^7heo> as soon as linux contains systemd 2014-11-13 15:24:41 [10:23] <^7heo> they will be linked together in source 2014-11-13 15:24:54 argument from false conspiracy-theorist-type premise :) 2014-11-13 15:25:04 <^7heo> whatever 2014-11-13 15:25:14 <^7heo> we don't comperehend each other 2014-11-13 15:25:17 <^7heo> but I don't give a fuck. 2014-11-13 15:25:47 no cross-platform apps will ever depend on systemd 2014-11-13 15:26:16 A lot of linux devs don't give a shit about being cross-platform 2014-11-13 15:26:19 no apps written by people without this idiotic "desktop linux" fdo.org mindset are likely to depend on systemd in critical ways (i.e. ways you can't just disable at compiletime) 2014-11-13 15:26:34 <^7heo> dalias: who the fuck cares about xplatform anymore? 2014-11-13 15:26:41 <^7heo> dalias: everything is "Linux" 2014-11-13 15:26:42 Also. Ok, an app may not hard depend on systemd in the sense that it may use devd on freebsd 2014-11-13 15:26:50 How does that help me, who has neither? 2014-11-13 15:26:57 ^7heo, the 2 gui apps that actually matter: firefox and chromium :-p 2014-11-13 15:27:00 <^7heo> where "Linux" means "GNU/Linux running systemd as PID 1" 2014-11-13 15:27:20 <^7heo> when Linux will include systemd in the source 2014-11-13 15:27:30 or do you actually want to run gnome-configurator-panel? :-p 2014-11-13 15:27:31 <^7heo> Google will have to use systemd or develop their own Kernel 2014-11-13 15:27:35 <^7heo> be it a linux fork 2014-11-13 15:27:43 linux is not including systemd in the source 2014-11-13 15:27:47 <^7heo> SteamOS will also have to run it 2014-11-13 15:27:48 that notion is utterly nonsensical 2014-11-13 15:27:56 <^7heo> why? 2014-11-13 15:27:57 and saying it just makes you look like a nutjob 2014-11-13 15:28:04 <^7heo> I don't care. 2014-11-13 15:28:06 and precludes anyone taking your arguments seriously 2014-11-13 15:28:08 <^7heo> I really don't give a fuck. 2014-11-13 15:28:15 <^7heo> I don't sleep enough anyway. 2014-11-13 15:28:19 i'm trying to be constructive here 2014-11-13 15:28:20 PLEASE 2014-11-13 15:28:24 <^7heo> how could I make sense when I can't think? 2014-11-13 15:28:24 if you want to fight against systemd 2014-11-13 15:28:35 <^7heo> I don't necessarily want to fight the inevitable. 2014-11-13 15:28:45 don't make anti-systemd look like a bunch of wackos 2014-11-13 15:28:49 <^7heo> like, I'm not fighting to stay alive: everybody dies, in the end. 2014-11-13 15:29:00 there are plenty of legitimate arguments against it that everyone can agree upon 2014-11-13 15:29:08 <^7heo> otherwise 2014-11-13 15:29:21 but too many divisive people doing stupid stuff and actually _helping_ systemd take over 2014-11-13 15:29:23 <^7heo> you're saying that systemd isn't going to be integrated in the linux kernel, right? 2014-11-13 15:29:26 dalias, except for the vocal part of the pro-systemd crowd 2014-11-13 15:29:35 <^7heo> I mean, source wise. 2014-11-13 15:29:42 There are no faults in systemd, according to those people :D 2014-11-13 15:29:55 None. Zero. Zilch. Nada. 2014-11-13 15:30:12 <^7heo> dalias: can you please confirm what I say before it is non-obvious that you are answering that? 2014-11-13 15:30:19 <^7heo> or should I repost so you can directly answer "yes"? 2014-11-13 15:30:20 (e.g. the debian MRA crowd who are fighting systemd just because some of the pro-systemd debian ppl are feminists) 2014-11-13 15:30:46 ^7heo, of course it's not going to be integrated into the kernel 2014-11-13 15:30:49 <^7heo> 'cause I want to point out, when it will be integrated in Linux, what you said, to you, in the logs. 2014-11-13 15:30:51 what on earth gave you that idea? 2014-11-13 15:30:58 <^7heo> dalias: 'cause it WILL happen. 2014-11-13 15:31:01 no it won't 2014-11-13 15:31:04 linux doesn't work that way 2014-11-13 15:31:07 <^7heo> oh yeah it will. 2014-11-13 15:31:13 <^7heo> if something bad can happen 2014-11-13 15:31:14 <^7heo> it will. 2014-11-13 15:31:19 <^7heo> if it's worse, it will happen faster. 2014-11-13 15:31:23 there have been hundreds of things that actually DO NEED TO BE in the kernel that linus refuses to put there and makes ppl do in userspace 2014-11-13 15:31:27 <^7heo> especially if it makes no sense. 2014-11-13 15:31:38 <^7heo> yeah 2014-11-13 15:31:43 <^7heo> but there's reason and money. 2014-11-13 15:31:49 he has a hard, often irrational, policy of never allowing things in the kernel if there's any possible way to do them in userspace 2014-11-13 15:31:50 <^7heo> and I believe that Linus isn't interested in the former. 2014-11-13 15:32:00 there is no money involved here 2014-11-13 15:32:06 <^7heo> of course not. 2014-11-13 15:32:10 and linus has plenty anyway 2014-11-13 15:32:15 dalias, if there's a reasonable way to do them in userspace 2014-11-13 15:32:16 <^7heo> Linus does not even want money. 2014-11-13 15:32:27 openfbtd, no, even when there's not a reasonable way to do them in userspace. 2014-11-13 15:32:27 As in, you can do a lot of drivers in userspace 2014-11-13 15:32:29 <^7heo> anyway 2014-11-13 15:32:36 <^7heo> how long are those logs kept? 2014-11-13 15:32:42 But we still put drivers into the kernel usually 2014-11-13 15:32:49 ^7heo, I keep logs forever :3 2014-11-13 15:32:50 openfbtd, think video drivers. obviously they belong in the kernel, but we're stuck with them in userspace :( 2014-11-13 15:33:02 <^7heo> openfbtd: then please stay in touch. 2014-11-13 15:33:02 dalias, eeeh, that's because of DRM 2014-11-13 15:33:22 <^7heo> 'cause WHEN (and not IF) systemd is merged into Linux "for consistency reasons", I'll point out this very log to dalias 2014-11-13 15:33:36 ^7heo, you can stop repeating yourself 2014-11-13 15:33:46 <^7heo> dalias: it's to make it easier to find it back in the logs. 2014-11-13 15:33:49 i am very happy being quoted anywhere that systemd will never be in the linux kernel 2014-11-13 15:34:04 <^7heo> I would like that too. 2014-11-13 15:34:04 Well. You both are on my dedi's raid forever 2014-11-13 15:34:14 Unless it fails 2014-11-13 15:34:14 <^7heo> =] 2014-11-13 15:37:27 <^7heo> Anyway, sorry dalias if you think that I contribute making the anti-systemd like a bunch of wackos; but I don't really consider the community to be making sense or having any logic. 2014-11-13 15:38:17 <^7heo> if they were replaced by a bunch of drunk sea lions, I bet that nobody would notice anything. 2014-11-13 15:44:18 systemd is a really horrible design and horrible idea, and as such, it's going to be eventually recognized as such 2014-11-13 15:44:47 and when that happens, the distros that have got its tentacles wrapped around all their stuff are going to be struggling to find an exit strategy 2014-11-13 15:45:30 <^7heo> dalias: you're far too optimist. 2014-11-13 15:45:46 no, just realist 2014-11-13 15:45:53 <^7heo> but I hope you prove me wrong. 2014-11-13 15:46:00 <^7heo> I just don't count on that. 2014-11-13 15:46:08 the only easy solution to this is heavy refactoring of systemd, or preferably reimplementation of the same interfaces, so that stuff can keep working while you remove all the breakage of systemd 2014-11-13 15:46:23 so that's probably the way it will happen when it does 2014-11-13 15:46:27 <^7heo> yeah 2014-11-13 15:46:30 <^7heo> and THAT will happen 2014-11-13 15:46:33 <^7heo> in the next 10 years. 2014-11-13 15:46:41 i'd expect the next 3-5 2014-11-13 15:46:47 <^7heo> yeah maybe 2014-11-13 15:46:53 the lifetime for these kinds of projects is usually even shorter than that 2014-11-13 15:46:53 <^7heo> on that we agree. 2014-11-13 15:47:04 think hal/policykit/consolekit/polkit/... 2014-11-13 15:47:14 <^7heo> it's going to be "the brick between linux and the users" 2014-11-13 15:47:18 they keep coming up with new broken replacements for the old one 2014-11-13 15:47:28 <^7heo> and it got MUCH more traction than the previous projects. 2014-11-13 15:47:31 i suspect the same will happen with systemd 2014-11-13 15:47:32 <^7heo> it's nothing comparable. 2014-11-13 15:47:43 <^7heo> I expect systemd to be different. 2014-11-13 15:47:58 <^7heo> people don't care about software, they care about religion. 2014-11-13 15:48:05 <^7heo> and systemd has an embedded religion. 2014-11-13 15:48:08 i think systemd has less traction and more resistence. it's just so much bigger in scope and impact that people actually notice it 2014-11-13 15:48:13 <^7heo> that's what the other software were missing. 2014-11-13 15:48:26 <^7heo> look at apple stuff 2014-11-13 15:48:35 <^7heo> it's totally religious 2014-11-13 15:48:48 <^7heo> there's no valid reason to get any Apple device from a reason-based point of view. 2014-11-13 15:48:57 i don't know anybody personally who's pro-systemd on religious grounds 2014-11-13 15:49:14 i suspect there are some from the desktop-linux scene 2014-11-13 15:49:41 but every instance i know of pro-systemd sentiment is basically just anti-sysvinit-sentiment 2014-11-13 15:49:44 <^7heo> unless you can quote facts 2014-11-13 15:49:45 <^7heo> and numbers 2014-11-13 15:49:51 <^7heo> or mathematical proofs 2014-11-13 15:49:56 <^7heo> everything is religious. 2014-11-13 15:49:59 <^7heo> like 2014-11-13 15:50:02 <^7heo> C is better than C++ 2014-11-13 15:50:03 <^7heo> it's religious. 2014-11-13 15:50:09 <^7heo> then you happen to produce tests. 2014-11-13 15:50:13 <^7heo> then it's not anymore. 2014-11-13 15:50:14 no, i have a proof :) 2014-11-13 15:50:18 <^7heo> ah? can I see it? 2014-11-13 15:50:33 sure, i just havent written it down yet 2014-11-13 15:50:41 <^7heo> then write it down. 2014-11-13 15:50:44 "the old way of doing stuff" is horribly bad, so an alternative that fixes that badness (which is known and understood) is attractive (even if it introduces new badness, since that new badness is not known and understood) 2014-11-13 15:50:49 <^7heo> unless you can produce it, it's religious. 2014-11-13 15:50:57 nsz, does it not fit in the margins, perchance? :-) 2014-11-13 15:50:57 <^7heo> and same for static vs dynamic. 2014-11-13 15:51:27 <^7heo> it's religious unless someone COMES with the same set of software, compiled statically and dynamically, and compares them. 2014-11-13 15:51:37 <^7heo> with facts to back up the claims. 2014-11-13 15:51:45 <^7heo> that said, I don't care if I sound religious about systemd. 2014-11-13 15:51:47 yes it needs a large margin.. but i guess i can use the margin of iso c++, it has plenty space 2014-11-13 15:51:54 "X is better than Y" is 'religious' in your sense of 'religious' 2014-11-13 15:51:58 <^7heo> it's far too complicated imho 2014-11-13 15:52:04 <^7heo> I'm not going to try to understand systemd. 2014-11-13 15:52:13 <^7heo> I might be interested in the problem it solves, tho. 2014-11-13 15:52:16 but "X does better than Y in terms of property Z" is usually not religious 2014-11-13 15:52:34 <^7heo> that is religious too. 2014-11-13 15:52:51 <^7heo> and if you will, the definition of maths is religious, but I'll stand with it for now. 2014-11-13 15:52:53 "static linking uses less memory (measured by commit charge or dirty pages) than dynamic linking" is objectively measurable 2014-11-13 15:53:09 <^7heo> dalias: I'm not sure it's true, so... 2014-11-13 15:53:17 <^7heo> because on a larger scale, it might use more memory. 2014-11-13 15:53:24 <^7heo> depending on many factors. 2014-11-13 15:53:25 read the "measured by" 2014-11-13 15:53:42 <^7heo> mainly how it is compiled and measured yes. 2014-11-13 15:53:43 .text size does not contribute anything to either of those 2014-11-13 15:56:07 <^7heo> but .text changes the RAM usage. 2014-11-13 15:56:18 only if use use it 2014-11-13 15:56:31 <^7heo> what? 2014-11-13 15:56:48 libc has lot of .text but i doubt you use long double _Complex function much 2014-11-13 15:56:58 :) 2014-11-13 15:57:10 so even if libc.so is large most of it wont get into the ram 2014-11-13 15:57:16 nsz, those might be bad examples because they're unlikely to be linked in a static program (unless you're using glibc ;-) 2014-11-13 15:57:21 <^7heo> so you're saying that the loader maps section of code in RAM *ONLY* when they are called? 2014-11-13 15:57:33 nsz, ^7heo is concerned about duplicate .text in lots of static programs 2014-11-13 15:57:39 ^7heo, yes, exactly 2014-11-13 15:57:39 ah 2014-11-13 15:57:49 <^7heo> I wasn't aware about that. 2014-11-13 15:57:51 the kernel faults in the pages only when they're accessed 2014-11-13 15:57:53 <^7heo> but it makes sense actually. 2014-11-13 15:57:55 they're always virtually mapped 2014-11-13 15:58:09 <^7heo> why are they virtually mapped? 2014-11-13 15:58:12 <^7heo> is there a reason for that? 2014-11-13 15:58:19 ^7heo, because they need to have an address 2014-11-13 15:58:21 <^7heo> aware of that I meant 2014-11-13 15:58:27 you can't reference something that doesn't have an address 2014-11-13 15:58:32 <^7heo> dalias: so then you can reuse that address? 2014-11-13 15:58:54 <^7heo> like, program starts, you map everything against a virtual memory 2014-11-13 15:58:59 <^7heo> and when functions get called 2014-11-13 15:59:02 <^7heo> you actually load them 2014-11-13 15:59:10 ^7heo, for 32-bit archs the size of virtual address space may be a consideration 2014-11-13 15:59:11 <^7heo> so if you map something and never use it 2014-11-13 15:59:16 <^7heo> it's never mapped 2014-11-13 15:59:26 for 64-bit it's irrelevant. you'll never use the whole address space anyway so it doesn't matter 2014-11-13 15:59:30 <^7heo> (first "map" means: in your code, second, in memory) 2014-11-13 15:59:45 <^7heo> right 2014-11-13 15:59:48 but dynamic linking uses more virtual address space too (because the whole .so is mapped even if you only use one function) 2014-11-13 15:59:56 <^7heo> yeah 2014-11-13 16:00:07 dalias: except large mmapped database 2014-11-13 16:00:08 whereas in static linking, functions that are unreachable may not be in the program image at all (and thus don't take up addresses) 2014-11-13 16:00:12 <^7heo> so the whole .so is loaded into RAM, right?\ 2014-11-13 16:00:21 that can use up the address space 2014-11-13 16:00:21 <^7heo> even if you don't need more than one call in it 2014-11-13 16:00:28 ^7heo, not loaded into ram. it just has a mapping in the process's virtual address space 2014-11-13 16:00:47 only parts of the .so that are actually accessed are loaded into ram (page fault on the first access, and the kernel loads them) 2014-11-13 16:00:55 <^7heo> dalias: but what if multiple programs call different functions from the same so then? 2014-11-13 16:01:40 <^7heo> they will all generate page faults? 2014-11-13 16:01:59 <^7heo> and therefore cause the loading of each function at every new call? 2014-11-13 16:02:38 if a given page of the .so is already loaded into memory due to one process using it, then it will be available to other processes using it without any page fault 2014-11-13 16:02:42 because of shared memory 2014-11-13 16:02:48 <^7heo> yes 2014-11-13 16:02:49 <^7heo> that I know. 2014-11-13 16:02:58 <^7heo> I'm concerned about the first call 2014-11-13 16:03:50 <^7heo> but I think I got it 2014-11-13 16:17:36 A new linux distro: no dbus, no udev. no NetworkManager, or systemd, or insserv. no python or perl or php. no X, no why, no zsh. no choice, no sleep and lots and lots of tweaking 2014-11-13 16:18:14 :-p 2014-11-13 16:18:39 +1 2014-11-13 16:18:39 we'll call it 90s Linux 2014-11-13 16:19:27 retrolinux? 2014-11-13 16:19:42 that prolly already exists 2014-11-13 16:20:51 what's wrong with sleep(1)? 2014-11-13 16:21:45 you can't tweak when you're sleeping. You can sleep when you're dead. 2014-11-13 16:22:47 fortunately you dont have to tweak much on a system that does not try to guess your network settings and other preferences 2014-11-13 16:24:28 i think it's a false dichotomy 2014-11-13 16:25:21 you generally don't want the system "guessing" your network settings except perhaps as a last-resort to find a connection to vpn-tunnel though, but you do want it reliably reproducing the settings you want under different conditions without painful manual intervention each time 2014-11-13 16:25:33 and NetworkManager horribly fails to do that 2014-11-13 16:25:46 it can't even manage priorities for access points 2014-11-13 16:26:04 much less different rules about which ones need mandatory vpn, or scripted bypass of login page, etc. to be usable 2014-11-13 16:39:40 <^7heo> does anyone know if there is a SOCKS proxy in alpine? 2014-11-13 16:40:02 <^7heo> 'cause there's no nc -X or -x flags 2014-11-13 16:41:23 you're probably using busybox nc. you could install the 'real' nc if there's a package for it 2014-11-13 16:41:37 personally the only socks proxy i've ever used is ssh 2014-11-13 16:42:32 <^7heo> dalias: the ssh client cannot connect through a SOCKS proxy, can it? 2014-11-13 16:43:07 no idea; i've never needed to do that 2014-11-13 16:43:37 <^7heo> thanks for reminding me that nc was a link to busybox 2014-11-13 16:43:49 <^7heo> I've installed the openbsd nc and now it works perfectly. 2014-11-13 16:45:25 i never understood netcat.. there are several different versions and all of them are borken.. in incompatible ways, but they are all called netcat 2014-11-13 16:45:50 the old 'real' one i used to use was written by somebody with no understanding of sockets 2014-11-13 16:45:50 so i moved to use socat because that at least is consistent 2014-11-13 16:46:12 it didn't handle close right at all and did some hack of sleeping for an interval to "let the socket drain" or some idiocy 2014-11-13 16:46:33 yes shutdown semantics is usually broken 2014-11-13 17:00:00 <^7heo> nsz: netcat-openbsd seems consistent 2014-11-13 17:00:57 netcat-openbsd++ 2014-11-13 17:30:35 ^7heo: dalias was suggesting that he uses ssh as a socks proxy. 2014-11-13 17:31:31 ssh -D ( + tsocks is the client doesn't support a proxy) 2014-11-13 17:31:42 s/is/if 2014-11-13 17:34:25 when i've been in places where i wasn't comfortable with even non-https, non-authenticated traffic going over unencrypted lines, i used ssh -D as a socks/http proxy 2014-11-13 17:34:40 vpn is probably more practical for general usage tho 2014-11-13 18:32:40 i see eudev as dead on arrival at this point 2014-11-13 18:32:44 ^^^ explain 2014-11-13 18:33:30 you yourself even said so 2014-11-13 18:33:47 11:42:43 +blueness | fbt, eudev is ultimately a loosing battle 2014-11-13 18:34:00 i am just taking a more extreme interpretation, that nobody will do the necessary kdbus work to keep it alive 2014-11-13 18:34:46 kaniini, there may be other options at the time 2014-11-13 18:34:56 when i see the kdbus code i'll decide 2014-11-13 18:35:07 but i may be able to pull out that code and isolate it 2014-11-13 18:35:25 when i said eudev is a loosing battle i meant it in the narrow sense of isolating it out of systemd 2014-11-13 18:35:42 i didn't mean there aren't other approaches that may be viable at the time 2014-11-13 18:35:59 its a loosing battle, but not a loosing war :) 2014-11-13 18:36:52 the deepest thing i'm looking for is to keep systemd off my system and continue using openrc 2014-11-13 21:18:11 <^7heo> ScrumpyJack: I know very well how to use a ssh proxy. 2014-11-13 21:18:24 <^7heo> the problem is how to proxy SSH thru a socks proxy. 2014-11-13 21:18:42 <^7heo> and THAT is not supported with the busybox nc implementation. 2014-11-13 21:19:46 kaniini, i don't see why any kdbus work would be needed for eudev 2014-11-13 21:19:57 the kernel is not going to remove the netlink-based interface for udev 2014-11-13 21:20:05 it would break the ability to run old dists with new kernels, etc. 2014-11-13 21:20:26 last i checked the kernel even still has a compatibility interface for ipchains and maybe even ipfwadm :-p 2014-11-13 21:24:48 hmm 2014-11-13 21:25:34 ^7heo: using ProxyCommand 2014-11-13 21:25:45 you'll need actual socks client doing the handshake iirc 2014-11-13 22:35:35 <^7heo> katlogic: I did that 7 hours ago, but thanks for the tip ;) 2014-11-13 22:35:52 <^7heo> katlogic: and guess what socks client you can use? nc. 2014-11-13 22:36:16 <^7heo> which brings us back to: <^7heo> nsz: netcat-openbsd seems consistent 2014-11-13 22:36:42 <^7heo> (around 1800 GMT+1) 2014-11-13 22:37:05 <^7heo> Anyhow 2014-11-13 22:37:07 <^7heo> right now 2014-11-13 22:37:10 <^7heo> crashing 2014-11-13 22:37:11 <^7heo> o/. 2014-11-13 23:06:50 dalias, the problem is the apps, not udev itself. As in, there will be apps that only use the new kdbus interface 2014-11-14 08:26:20 morning 2014-11-14 08:31:07 morning 2014-11-14 08:33:11 moin 2014-11-14 08:38:26 all this systemd talk i find quite depressing. 2014-11-14 08:39:44 why is that? 2014-11-14 08:40:24 i don't want/need/like systemd - but it seems inevitable 2014-11-14 08:41:00 then you should be glad so many ppl are talking about it. 2014-11-14 09:12:43 to be honest i'm a bit lost, when it comes to behind-the-scene of systemd, at least when i read stuff like that: http://seclists.org/oss-sec/2014/q4/609 2014-11-14 09:49:03 ScrumpyJack: avoiding it pretty fine so far. i mean there's some centos boxes I'm working on that have it 2014-11-14 09:49:13 but i dont care about those boxes either 2014-11-14 09:49:34 journalctl is same stupid as alog on aix was 2014-11-14 09:49:40 don't even think of it 2014-11-14 09:52:27 It's even possible to roll distros that ship systemd in base without it 2014-11-14 09:52:33 With varying success. 2014-11-14 09:53:39 Depending on the distro 2014-11-14 09:53:57 I gave up with debian, but Arch is easy to rip systemd out of. 2014-11-14 09:56:05 i'm still on debian 7, so no systemd here yet. I didnt know it was easy to remove systemd from Arch, but i'm moving to alpine anyway 2014-11-14 09:56:36 (I mean i'm moving from arch to alpine) 2014-11-14 09:58:52 ScrumpyJack, https://fleshless.org/pages/spark.html as one of the possible configs 2014-11-14 09:59:29 yes, i was looking at that while trying to be friendly :) 2014-11-14 10:00:05 Are you mad? This is IRC. We yell at people. 2014-11-14 10:14:37 <^7heo> moin moin 2014-11-14 10:40:47 <^7heo> damn, I have an .so missing when I 'apk add', and if I create a symlink for that .so, it doesn't solve the problem. 2014-11-14 10:49:28 <^7heo> okay, apk update fixed the issue. 2014-11-14 10:49:31 <^7heo> sorry for the noise. 2014-11-14 10:49:43 <^7heo> now I just wonder if I can safely rm the .so 2014-11-14 10:51:35 <^7heo> I guess that apk info -L can answer that question 2014-11-14 10:53:01 <^7heo> yep it does. 2014-11-17 08:17:52 morning 2014-11-17 08:20:02 moin 2014-11-17 09:14:15 <^7heo> moin everyone 2014-11-17 09:14:29 morning! 2014-11-17 09:14:40 <^7heo> hey ginjachris ;) 2014-11-17 11:35:58 <^7heo> is it me or is git missing the "interactive" feature in alpine? 2014-11-17 11:36:07 <^7heo> $ git add --interactive 2014-11-17 11:36:08 <^7heo> git: 'add--interactive' is not a git command. See 'git --help'. 2014-11-17 11:39:21 seems that there's a problem with that space between add and -- 2014-11-17 11:39:48 what about git add -i ? 2014-11-17 11:40:14 Is your git add an alias ? 2014-11-17 11:41:37 <^7heo> quinq: there is no problem in the command I input. 2014-11-17 11:41:42 <^7heo> as you can read. 2014-11-17 11:41:56 <^7heo> and no, I have no aliases defined. 2014-11-17 11:42:05 Well, I do not trust blindy everything that is written en IRC 2014-11-17 11:42:07 <^7heo> I would be interested to know how to define aliases in ash, btw. 2014-11-17 11:42:16 <^7heo> quinq: not written, pasted. 2014-11-17 11:42:18 Don't get angry, please 2014-11-17 11:42:29 Just trying to help 2014-11-17 11:42:31 <^7heo> quinq: don't assume I am angry, please :) 2014-11-17 11:42:42 Ok, I have things to do 2014-11-17 11:42:54 <^7heo> fine by me. :) 2014-11-17 11:43:26 <^7heo> and, btw, git add -i does return, exactly like git add -a, the same line as git add --interactive. 2014-11-17 11:43:42 <^7heo> sorry, I meant git add -p, not git add -a 2014-11-17 12:26:33 ^7heo: apk add git-perl 2014-11-17 12:27:14 ^7heo: there are some of the git stuff that is implemented in perl 2014-11-17 12:27:18 <^7heo> ncopa: thanks a bunch 2014-11-17 12:27:30 i didnt want pull in perl as a dep, since its not always wanted 2014-11-17 12:27:40 <^7heo> Oh, now I get an error. 2014-11-17 12:27:53 <^7heo> can't locate Git.pm... 2014-11-17 12:27:57 oh 2014-11-17 12:27:59 thats a bug 2014-11-17 12:28:13 <^7heo> should I do something? 2014-11-17 12:28:14 <^7heo> report? 2014-11-17 12:28:15 <^7heo> fix? 2014-11-17 12:28:23 i'll fix it immediatly 2014-11-17 12:28:29 <^7heo> many thanks. 2014-11-17 12:29:04 <^7heo> btw, is it worth it to wait for upstream to include the changes I submit + do a release, or should I submit my apk with a patch? 2014-11-17 12:30:08 if you can wait 4-5 mins... 2014-11-17 12:30:26 i know whats wrong and how to fix it 2014-11-17 12:30:32 and its compiling... 2014-11-17 12:31:10 <^7heo> I can wait. 2014-11-17 12:31:15 <^7heo> it's to submit the patch to upstream ;) 2014-11-17 12:31:32 its not an upstream bug 2014-11-17 12:31:37 <^7heo> not that one. 2014-11-17 12:31:47 :) 2014-11-17 12:31:53 <^7heo> I'm waiting for -p to work in order to be able to submit a patch for upstream for hhpc. 2014-11-17 12:31:54 sorry, i cannot work ant irc at same time :) 2014-11-17 12:32:00 <^7heo> which is a softwre I just packaged for alpine. 2014-11-17 12:32:34 <^7heo> but it's not yet on the ML because I want my change to be pulled in upstream, and a release to happen, so I can do it in a much simpler way for alpine. 2014-11-17 12:32:50 i pushed git-perl fix 2014-11-17 12:33:05 <^7heo> all that because the guy didn't write an "install" target in his Makefile... 2014-11-17 12:33:14 its done in, 5 2014-11-17 12:33:16 4 2014-11-17 12:33:18 3 2014-11-17 12:33:18 <^7heo> :D 2014-11-17 12:33:21 2 2014-11-17 12:33:27 1.5 2014-11-17 12:33:31 1.4 2014-11-17 12:33:33 <^7heo> huhu 2014-11-17 12:33:38 1.3 2014-11-17 12:33:40 1 2014-11-17 12:33:41 0 2014-11-17 12:33:44 <^7heo> \o/ 2014-11-17 12:33:58 <^7heo> hmm 23 packages to upgrade... 2014-11-17 12:34:11 <^7heo> Well, okay. 2014-11-17 12:34:57 <^7heo> alpine works so great as a desktop... 2014-11-17 12:35:51 <^7heo> wow, that's weird. 2014-11-17 12:35:58 <^7heo> now if I git add -p I get the colors 2014-11-17 12:36:40 <^7heo> but git diff displays the color codes.... 2014-11-17 12:36:59 <^7heo> does it come from less? 2014-11-17 12:37:41 <^7heo> yeah, it does... I wonder how to make it use something else than busybox's less... 2014-11-17 12:38:19 <^7heo> works with normal less. 2014-11-17 12:38:21 ^7heo: export PAGER= 2014-11-17 12:38:30 what i normally do: apk add less 2014-11-17 12:38:32 <^7heo> yeah, but less wasn't installed, it was busybox's one. 2014-11-17 12:38:42 <^7heo> and installing it solved the issue ;) 2014-11-17 12:38:52 there was suggestion to add color support to busybox less some time ago 2014-11-17 12:38:52 <^7heo> but thanks for the PAGER var, it's worth it to know. 2014-11-17 12:39:15 <^7heo> yeah, and also detection of the length of the buffer, so it doesn't invoke less if it's less than the term size... 2014-11-17 12:39:21 <^7heo> would be fun too. 2014-11-17 12:39:26 but unfortunally, the patch submitter mentioned systemd together with the patch to it got rejected :) 2014-11-17 12:39:37 <^7heo> wtf?! 2014-11-17 12:39:56 <^7heo> like, people submit patches with insults in it now?! 2014-11-17 12:40:01 <^7heo> Oo 2014-11-17 12:40:26 <^7heo> btw, many thanks for the git fix 2014-11-17 12:40:29 <^7heo> it works perfectly now. 2014-11-17 12:42:40 http://lists.busybox.net/pipermail/busybox/2014-January/080353.html 2014-11-17 12:44:20 <^7heo> v_v 2014-11-17 12:45:23 <^7heo> I agree with the answer, btw. 2014-11-17 12:48:17 i think it'd be nice to have colors in bb less, due to git 2014-11-17 12:48:44 <^7heo> yeah 2014-11-17 12:48:50 <^7heo> same here. 2014-11-17 12:48:51 if he would have said: "i like this due to make git log lookg nice" 2014-11-17 12:48:59 then it might have been accepted 2014-11-17 12:49:02 <^7heo> yeah 2014-11-17 15:10:36 one can also use http://tasktools.org/projects/clog.html to color the git output 2014-11-17 15:10:41 maybe git log --no-color 2014-11-17 15:11:09 export PAGER="clog --file ../.clogrc git|less" 2014-11-17 15:12:09 wonder if de-coupling the color producing code in git would be an worth effort :) 2014-11-17 15:21:25 you can config git to not do color 2014-11-17 15:21:44 <^7heo> yeah, but that is kind of missing the point. 2014-11-17 15:23:13 <^7heo> but I fear that vkrishn has missed the main idea. git log operates in color by default... 2014-11-17 15:25:23 Colour! That's the work of the devil! Much like singing, dancing and cavorting! :p 2014-11-17 15:26:49 on a more serious note: I think I'd just leave it with colour :) 2014-11-17 15:27:08 <^7heo> ginjachris: technically, it's the work of your brain. 2014-11-17 15:27:23 right, pass me that hand drill! 2014-11-17 15:28:15 have you seen the film 'Pi', he sees eveything in black and white. Weird film. Good, but weird. 2014-11-17 15:28:59 <^7heo> yeah, but by default, on alpine, git diff looks like: http://sprunge.us/AFGa 2014-11-17 15:29:10 <^7heo> nah I didn't. 2014-11-17 15:29:25 <^7heo> He sees everything in black and white because he is a tiger, also. 2014-11-17 15:30:07 i am a tiger, hear me roar! 2014-11-17 15:30:11 <^7heo> :P 2014-11-17 15:30:17 (sorry, in a weird mood this afternoon) 2014-11-17 15:30:24 <^7heo> don't worry 2014-11-17 15:30:25 ginjachris: i once had a roomie who summarized Pi with this: 2014-11-17 15:30:26 <^7heo> it's life 2014-11-17 15:30:34 "woah, does that guy never stop thinking" 2014-11-17 15:30:47 it made me a little envious 2014-11-17 15:31:07 lol yeah, I see what you mean 2014-11-17 15:31:20 I'm pretty sure he didn't do so much thinking at the end of the film 2014-11-17 15:35:36 right, time for a tea 2014-11-17 15:42:45 <^7heo> who does have the rights to edit the web page? 2014-11-17 15:43:52 <^7heo> speficically that one: http://alpinelinux.org/about 2014-11-17 16:28:59 i don't have a usb install to check, so hoping someone here could check. base apks are in /apks/x86_64. does an apk upgrade of a usb stick upgrade the packages in /apks/x86_64 or does it add the packages to APKINDEX? 2014-11-17 16:38:40 ScrumpyJack: i *think* it puts them in APKINDEX 2014-11-17 16:38:51 I'm afriad that I must depart 2014-11-17 16:38:57 Farewall, Comrades! 2014-11-17 16:40:51 ScrumpyJack: here 'apk upgrade' upgrades /media/usb/cache packages 2014-11-17 16:41:30 aha 2014-11-17 16:41:30 ScrumpyJack: not /media/usb//apks/x86_64 2014-11-17 16:42:30 ScrumpyJack: and only packages that you have load with 'apk add' 2014-11-17 16:42:33 so those in cache/packages squash those in apks/x86_64 perhaps 2014-11-17 16:43:21 yes 2014-11-17 16:44:59 apks/x86_64 packages came with the iso 2014-11-17 16:45:31 ah i see. no i mean if you apk update and akp upgrade and it upgrades a the curl package. does that replace the curl package from the base image in apks/x86_64 that came with the iso? 2014-11-17 16:45:45 nope 2014-11-17 16:46:06 packages are not replaced 2014-11-17 16:46:22 hmm 2014-11-17 16:48:37 neither in '/media/usb/cache' 2014-11-17 16:50:11 for example, I have a lot of firefox versions on my '/media/usb/cache' 2014-11-17 17:35:31 re: git log operates in color, that's the point, I thought is what is level of complexity of implementing that color output 2014-11-17 17:35:57 if much than the effort to de-couple it is worth 2014-11-17 17:36:25 the* thought is, 2014-11-17 17:41:03 Does anybody know when/why go support was removed from gcc in Alpine? I see where ncopa added it about 3 years ago, not sure when it was removed. 2014-11-17 17:41:45 ^ referring to http://bugs.alpinelinux.org/issues/624 2014-11-17 17:47:27 kinda, standard linux util for coloring outputs 2014-11-17 18:14:21 blakes: from the gcc APKBUILD: "# Go needs {set,make,swap}context, unimplemented in musl" 2014-11-17 20:32:46 hello guys 2014-11-17 20:34:08 is this a rolling release distro? 2014-11-17 20:35:22 n-iCe: it can be if u follow edge 2014-11-17 20:37:31 Thank you AmatCoder 2014-11-17 23:30:15 still here pnutzh4x0r ? 2014-11-17 23:39:22 or anyone here? 2014-11-17 23:59:53 n-iCe: ? 2014-11-18 00:02:50 hi pnutzh4x0r 2014-11-18 00:03:12 hello 2014-11-18 00:03:20 lol 2014-11-18 00:04:49 hehe 2014-11-18 00:04:55 I am thinking in move to alpine 2014-11-18 00:05:52 it is based on what 2014-11-18 00:06:24 i think it had roots in gentoo... but has long moved away from that 2014-11-18 00:06:40 it's not really based on anything 2014-11-18 00:06:48 what's the download manager package name? 2014-11-18 00:06:51 it has a custom package system 2014-11-18 00:06:53 apk 2014-11-18 00:07:28 but if I want to isntall a package 2014-11-18 00:07:32 http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management 2014-11-18 00:07:35 can't I? like skype, chrome 2014-11-18 00:07:53 those probably wouldnt work b/c alpine uses musl libc 2014-11-18 00:08:00 rather than gnu libc 2014-11-18 00:08:04 oh, so this is not for me 2014-11-18 00:08:08 I need chrome for my chromecast 2014-11-18 00:08:10 so proprietary stuff wouldnt work 2014-11-18 00:08:16 chromium maybe? 2014-11-18 00:08:23 someone is working on a chromium build 2014-11-18 00:08:31 im not sure what the condition is 2014-11-18 00:08:55 oh 2014-11-18 00:09:00 well, still not useful for me then 2014-11-18 00:09:00 thanks! 2014-11-18 00:09:28 np 2014-11-18 00:09:36 gl on your search 2014-11-18 08:00:58 Good Morning all 2014-11-18 08:02:14 morning 2014-11-18 08:05:32 ginjachris, good evening 2014-11-18 08:06:05 :) 2014-11-18 08:57:25 <^7heo> moin ncopa 2014-11-18 08:57:52 morning 2014-11-18 09:03:16 <^7heo> damn, the developer of hhpc isn't answering... 2014-11-18 09:04:07 <^7heo> is it okay to submit two APKBUILDs, one for the current release (of a given software, not of alpine), and one for the next release + future releases? 2014-11-18 09:09:17 https://www.youtube.com/watch?v=QrGrOK8oZG8 2014-11-18 09:09:54 <^7heo> wtf is that? 2014-11-18 09:10:12 just watch :P 2014-11-18 09:10:32 <^7heo> I am 2014-11-18 09:10:36 <^7heo> but really... 2014-11-18 09:10:56 <^7heo> that's lame... 2014-11-18 09:11:10 keep watching :P 2014-11-18 09:11:25 <^7heo> I can't, I have to work at the same time ;) 2014-11-18 09:11:29 <^7heo> wait 2014-11-18 09:11:36 <^7heo> is that NSFW?! 2014-11-18 09:12:06 <^7heo> okaay I keep the link for home. 2014-11-18 09:13:10 leaning towards not NSFW 2014-11-18 09:20:24 Not Not Safe For Work 2014-11-18 09:20:27 NNSFW 2014-11-18 09:21:51 <^7heo> or, clearer: NTBLNSFW. 2014-11-18 09:22:04 <^7heo> (Not To Be Labelled Not Safe For Work) 2014-11-18 09:24:13 ^7heo, my b 2014-11-18 09:25:44 10:09 < ^7heo> wtf is that? 2014-11-18 09:25:46 indeed :) 2014-11-18 10:02:52 well there is Goth Fitness https://www.youtube.com/watch?v=4u29MoqY01k 2014-11-18 10:02:54 :P 2014-11-18 10:08:24 quinq, at least enjoyable? 2014-11-18 10:09:32 Very much ! 2014-11-18 10:09:47 Although I haven't watched it again since first time, it was good 2014-11-18 10:09:54 that was weird. 2014-11-18 10:10:07 I'll give it a go later, now is too soon 2014-11-18 15:26:12 <^7heo> ncopa: any idea why the flags for my docker are different than the ones I find in the doc? 2014-11-18 15:26:16 <^7heo> is it a version problem? 2014-11-18 15:26:42 falgs? 2014-11-18 15:26:48 flags? 2014-11-18 15:26:58 i have no idea. never used docker 2014-11-18 15:27:01 <^7heo> yeah, like 2014-11-18 15:27:06 <^7heo> the doc' says 2014-11-18 15:28:00 <^7heo> https://docs.docker.com/reference/run/#expose-incoming-ports 2014-11-18 15:28:02 <^7heo> for example 2014-11-18 15:28:08 <^7heo> and the man says 2014-11-18 15:28:19 <^7heo> -p="" 2014-11-18 15:28:20 <^7heo> Path to use for daemon PID file. Default is /var/run/docker.pid 2014-11-18 15:28:34 <^7heo> so it's very different... 2014-11-18 15:29:17 <^7heo> and apparently, it's not a version problem, 'cause the version in ubuntu is also 1.2.0 and the -p flag can be used like : to specify a port binding. 2014-11-18 15:50:37 unhelpful, i know, but i use lxc with great success 2014-11-18 15:51:37 i was tempted by docker 2014-11-18 15:54:04 <^7heo> docker is VERY simple to use. 2014-11-18 15:54:10 <^7heo> and it has a LOT of traction. 2014-11-18 15:54:45 i read that "Docker is an open platform for developing, shipping, and running applications". I hate that kind of talk. 2014-11-18 15:56:05 yes, very simple. i like that they have their own libs 2014-11-18 15:56:34 <^7heo> yeah, the talk is shit. 2014-11-18 15:56:38 <^7heo> but they have to adapt. 2014-11-18 15:57:06 <^7heo> if you want to "sell" anything, you have to lie and say bullshit :P 2014-11-18 15:57:16 i looked before version 1. i should look again 2014-11-18 15:59:27 <^7heo> it's incredibly easy and convenient. 2014-11-18 15:59:59 <^7heo> but the alpine version is rather different from the ubuntu version, for some reasons... 2014-11-18 16:00:40 "Why docker? We want your environment to work better." 2014-11-18 16:00:59 /facepalm 2014-11-18 16:01:55 <^7heo> yeah 2014-11-18 16:01:56 <^7heo> bullshit. 2014-11-18 16:02:14 are you using the apk in edge or have you rolled your own? 2014-11-18 16:02:16 <^7heo> but that's necessary for apes to get interested in it. 2014-11-18 16:02:20 <^7heo> apk in edge. 2014-11-18 16:02:52 maybe update the version in edge to docker 1.3? 2014-11-18 16:03:54 need to go see u 2014-11-18 16:05:01 i'm off home, or i'd take a look. 2014-11-18 16:05:25 good luck 2014-11-18 16:06:01 <^7heo> o/ 2014-11-18 16:09:39 <^7heo> ncopa: thanks for webkitgtk2 :) 2014-11-18 16:09:46 <^7heo> now one can build surf without too many problems. 2014-11-18 20:31:38 <^7heo> YAY, my change has been merged, I can just post my apkbuild as is :D 2014-11-18 20:31:41 <^7heo> \o/ 2014-11-19 01:26:24 Greetings! Is there a version of Alpine Linux for the MIPS architecture? I am interested in installing Alpine Linux onto a Ubiquiti Edgerouter Lite/Edgerouter Pro for NHRP. 2014-11-19 01:26:28 And DMVPN 2014-11-19 01:29:42 Sachiru: afaik no half-or-less official build 2014-11-19 01:29:52 there's a little bit for arm in terms of raspi 2014-11-19 01:29:53 Ah. 2014-11-19 01:30:00 Hmm. 2014-11-19 01:30:07 DMVPN would be nice to have 2014-11-19 01:30:31 very much so 2014-11-19 01:31:00 Sachiru: wait a few more hours until the main devs are around 2014-11-19 08:10:51 Good Morning everyone 2014-11-19 08:21:02 mornin 2014-11-19 10:43:05 I'm going to ask a question that may result in much verbal abuse/racous laughter/banishment....but here goes: 2014-11-19 10:43:15 <^7heo> moin ginjachris 2014-11-19 10:43:17 <^7heo> soup? 2014-11-19 10:43:38 Is it possible to get the raspberry pi releases in the form of a .img file? This would make one's life much easier 2014-11-19 10:43:50 and I want something reliable on my Pi :) 2014-11-19 10:43:56 <^7heo> ginjachris: you mean, a dd-able img file? 2014-11-19 10:44:09 worse....an Win32-imager .img file 2014-11-19 10:44:24 <^7heo> that's not really understandable... 2014-11-19 10:44:33 <^7heo> why win32-imager? 2014-11-19 10:45:13 Windoze at home....and telling it to just copy the .img over to the card is easier than me trying to do anything fancy, which usually results in disaster lol 2014-11-19 10:45:15 <^7heo> is it the default R-Pi stuff? 2014-11-19 10:45:22 <^7heo> nah but... 2014-11-19 10:45:38 <^7heo> using dd to dd an image to the card is no big deal. 2014-11-19 10:45:50 <^7heo> and if you have windows only, you can get dd for it... 2014-11-19 10:45:59 that's an idea 2014-11-19 10:46:04 <^7heo> ttp://www.chrysocome.net/dd 2014-11-19 10:46:11 but...is there an image? 2014-11-19 10:46:36 thanks ^7heo! 2014-11-19 10:46:50 <^7heo> also 2014-11-19 10:47:12 <^7heo> win32 disk imager advertises "Raw" disk image writing. 2014-11-19 10:47:21 <^7heo> which should technically be the exact same as dd, afaik. 2014-11-19 10:47:34 yes that sounds right 2014-11-19 10:47:38 <^7heo> now 2014-11-19 10:47:52 <^7heo> doing a release of an alpine disk as an img isn't too hard. 2014-11-19 10:48:03 <^7heo> just setup alpine in a vm 2014-11-19 10:48:16 <^7heo> do an img of your disk from the installer after installation 2014-11-19 10:48:20 <^7heo> save that img somewhere. 2014-11-19 10:48:22 <^7heo> done 2014-11-19 10:48:29 <^7heo> right? 2014-11-19 10:48:35 <^7heo> that process can be automatized easily. 2014-11-19 10:48:37 yep that makes sense 2014-11-19 10:48:55 <^7heo> but the point is that you need to prepare an install script if you want to automatize it. 2014-11-19 10:49:05 <^7heo> I can maybe try to get the image done over docker if you want. 2014-11-19 10:49:28 oh don't worry mate, I am sure youhave better things to do with your time! 2014-11-19 10:49:38 <^7heo> yeah, but that sounds fun :P 2014-11-19 10:50:02 <^7heo> brb 2014-11-19 10:50:08 ok, good luck! 2014-11-19 10:58:02 <^7heo> back 2014-11-19 11:03:00 <^7heo> okay, not sure if it's easy to do with docker, because docker isn't proper virtualization. 2014-11-19 11:03:07 <^7heo> but with kvm, it must be very easy. 2014-11-19 11:10:52 it's ok, I will try and do something when I get time 2014-11-19 11:10:58 time is a scare resource at the moment :( 2014-11-19 11:12:02 <^7heo> thing is 2014-11-19 11:12:07 <^7heo> the more you do, the more you can do. 2014-11-19 11:12:11 <^7heo> and not because you have more time. 2014-11-19 11:12:19 <^7heo> but because you know how to do stuff ;) 2014-11-19 11:12:24 <^7heo> anyhow 2014-11-19 11:12:51 indeed....I know stuff, but it's the stuff I don't know that takes up all my time 2014-11-19 11:13:13 <^7heo> exactly. 2014-11-19 11:13:23 and thanks for trying :) 2014-11-19 11:13:29 <^7heo> well 2014-11-19 11:14:13 <^7heo> what I can do is write scripts to minimize the work needed to do what you want. 2014-11-19 11:14:45 Oh don't worry about it; it's not worth your trouble! 2014-11-19 11:15:10 <^7heo> it is. 2014-11-19 11:15:45 <^7heo> contributing to Alpine, and finding the best way to solve a problem that could happen to more than one... 2014-11-19 11:15:55 <^7heo> that's worth the trouble. 2014-11-19 11:16:17 well, I'll try using DD to copy the files over first....that might do it 2014-11-19 11:16:47 <^7heo> nah but 2014-11-19 11:16:49 <^7heo> really 2014-11-19 11:17:12 <^7heo> it might not even need virtualization... 2014-11-19 11:18:17 no i don't think so....there is a raspberry pi alpine "image" available, but it's a tar.gz of files, not a .img file.....last time I did anything on the pi everything was .img files but that seems to have changed for some reason 2014-11-19 11:18:19 <^7heo> I mean, look at it this way: create a fs in a file, have a node in /dev created for that fs, and use the "setup-alpine" script on that /dev node... 2014-11-19 11:18:23 <^7heo> that might be just enough, right? 2014-11-19 11:18:35 <^7heo> and that way, you could setup anything you want, from an already installed alpine. 2014-11-19 11:19:08 yeah that could work. I really need to sit down and messa round with it 2014-11-19 11:20:18 <^7heo> I think it would even be possible to modify the setup-alpine script so it could take more parameters 2014-11-19 11:20:27 <^7heo> and totally skip the interactive bits. 2014-11-19 11:20:33 <^7heo> and create the install in a file directly. 2014-11-19 11:20:42 <^7heo> that way 2014-11-19 11:20:46 <^7heo> you just call, say: 2014-11-19 11:20:48 yeah setup-alpine can reference a file 2014-11-19 11:21:00 with preconfigured options 2014-11-19 11:21:19 problem here is just getting the pi image to sd card 2014-11-19 11:21:33 and it's gotta be the pi image of course, since that's ARM architecture 2014-11-19 11:21:44 <^7heo> setup-alpine -f rpi.0.0.1.img -t ext4 -k us -i 192.168.0.34/24 ... 2014-11-19 11:22:00 <^7heo> and then you would have rpi.0.0.1.img in the current directory 2014-11-19 11:22:21 <^7heo> ready to be dd 2014-11-19 11:36:58 <^7heo> apk is so awesome... 2014-11-19 11:42:14 yes it is 2014-11-19 12:07:17 <^7heo> okay, I need a proper fcgiwrap in C. 2014-11-19 12:07:28 <^7heo> I'm going to check the perl source to see if I can understand something :P 2014-11-19 12:08:03 <^7heo> wait, it's written in C, why is it always spawned in perl, then?! 2014-11-19 12:09:32 it is becoming..... 2014-11-19 12:09:36 ......self-aware! 2014-11-19 12:09:56 destroy it quickly, or we are all doomed! 2014-11-19 12:09:59 <^7heo> ? 2014-11-19 12:10:11 I'm just being silly :) 2014-11-19 12:10:14 <^7heo> I don't destroy stuff when I don't understand why it's silly :P 2014-11-19 12:10:22 <^7heo> however 2014-11-19 12:10:35 <^7heo> I wonder why autotools is needed for a one-file C project... 2014-11-19 12:10:37 <^7heo> really... 2014-11-19 12:10:51 <^7heo> fcgiwrap is a good idea, but implementation is poor. 2014-11-19 12:11:20 <^7heo> i.e. what's existing is relying on poor tools, and not all of what's needed is existing within the wrapper. 2014-11-19 12:11:28 <^7heo> I might wanna fork it and make it complete. 2014-11-19 12:11:41 <^7heo> and at the same time, solve the problems I encounter. 2014-11-19 12:11:45 <^7heo> let's check the source. 2014-11-19 12:25:17 ^7heo, suckless fcgiwrap? :D 2014-11-19 12:28:06 <^7heo> openfbtd: it's existing? 2014-11-19 12:28:29 No, I meant that you might make one! 2014-11-19 12:28:38 It would be pretty cool actually. 2014-11-19 12:29:32 <^7heo> openfbtd: lemme decypher the perl code first :P 2014-11-19 12:29:40 <^7heo> but yeah, I could make one. 2014-11-19 12:29:54 <^7heo> and it not something that would require too much maintainance I believe. 2014-11-19 12:31:34 <^7heo> it's incredible how easily people copy/paste glue-code rather than checking if it's easy to implement it in one of the components being glued... 2014-11-19 12:31:37 <^7heo> like 2014-11-19 12:31:40 <^7heo> https://gist.github.com/stran12/1394757 2014-11-19 12:31:51 <^7heo> this "spawn-fcgi" is super simple. 2014-11-19 12:32:18 <^7heo> and it could take place in the fcgiwrap code without too much effort. 2014-11-19 12:32:36 <^7heo> resulting in a much simpler setup. 2014-11-19 12:32:38 <^7heo> v_v 2014-11-19 12:33:02 <^7heo> ACTION is astonished to see how complicated things get 'cause of a few people reusing rather than criticizing. 2014-11-19 12:33:06 <^7heo> Now, let's contribute. 2014-11-19 12:34:01 <^7heo> okay, the C file is 845 lines. 2014-11-19 12:34:16 <^7heo> I hesitate between forking the project, or doing another one... 2014-11-19 12:34:54 <^7heo> I mean, it's probably worth a fork, but on the other hand... I could simply rewrite it in not too much time, allowing me to release it under BSD license. 2014-11-19 12:35:52 <^7heo> okay, I opt for a clean rewrite. 2014-11-19 12:36:08 <^7heo> the presence of systemd-related code did help me to chose ;P 2014-11-19 12:44:10 <^7heo> I am tempted to call that cgid. What do you think, openfbtd? 2014-11-19 12:44:17 <^7heo> or scgid? 2014-11-19 12:44:47 <^7heo> (s for suckless ofc) 2014-11-19 12:53:52 <^7heo> since scgi exists, I think that it's a bad idea to call it that way. 2014-11-19 13:05:16 Why cgid? 2014-11-19 13:08:34 <^7heo> to satire the use of 'd' at the end of everything as a hype thing. 2014-11-19 13:08:44 <^7heo> but more seriously, because it's going to be a daemon running CGIs. 2014-11-19 13:21:25 ^7heo, oh. cgi daemon. Makes sense 2014-11-19 13:21:51 Could even be fcgid 2014-11-19 13:22:10 As its purpose is to wrap cgi into fcgi :D 2014-11-19 13:22:13 <^7heo> no, 'cause I want to support both FCGI and SCGI. 2014-11-19 13:22:16 Oh 2014-11-19 13:22:18 <^7heo> and 2014-11-19 13:22:25 Well, cgid is a good name then 2014-11-19 13:22:31 <^7heo> it daemonizes CGIs, not FCGIs or SCGIs. 2014-11-19 13:22:33 <^7heo> like 2014-11-19 13:22:55 <^7heo> d daemonizes , not the result of the daemonization. 2014-11-19 13:23:08 <^7heo> that's what's bugging me with fcgiwrap 2014-11-19 13:23:14 <^7heo> it should be named cgiwrap 2014-11-19 13:23:21 <^7heo> it's not wrapping ANY fcgi ever. 2014-11-19 13:23:26 True. 2014-11-19 13:23:35 <^7heo> it's wrapping CGIs. Into FCGIs, okay, but that shouldn't be said that way. 2014-11-19 15:23:52 afternoon 2014-11-19 15:25:13 good afternoon 2014-11-19 15:26:35 <^7heo> moin. 2014-11-19 15:33:33 sles12 comes with systemd and wicked. Boo 2014-11-19 15:35:40 <^7heo> what did you expect? 2014-11-19 15:35:47 <^7heo> sles is even worse than noobuntu. 2014-11-19 15:39:30 what doesnt run systemd these days? :p 2014-11-19 15:39:46 <^7heo> alpine 2014-11-19 15:40:08 i'm in an enterprise environment here 2014-11-19 15:40:54 <^7heo> ScrumpyJack: oh, that changes everything. Sorry, I didn't know. 2014-11-19 15:41:10 we are on centos 6 here 2014-11-19 15:41:13 so no systemd yet 2014-11-19 15:41:22 <^7heo> yeah, makes sense. 2014-11-19 15:41:27 <^7heo> gotta keep your jobs, right? 2014-11-19 15:41:55 the things we put up with for money... 2014-11-19 15:42:35 <^7heo> I wish you the best. 2014-11-19 15:42:36 <^7heo> really. 2014-11-19 15:43:31 meh, i just help run a HPC cluster 2014-11-19 15:43:38 so once things are setup we leave it alone 2014-11-19 15:43:46 most of the work is helping users debug their applications 2014-11-19 15:43:59 <^7heo> that I know. 2014-11-19 15:44:01 and fighting strange combinations of software 2014-11-19 15:44:22 like openmpi installed w/ x, y, z compilers with foo bar networking libraries 2014-11-19 15:45:06 we are on sles9 :) 2014-11-19 15:49:52 Does wicked depend on systemd? 2014-11-19 15:50:22 <^7heo> of course, would it be wicked if not? 2014-11-19 16:50:03 bye bye all! Have a great evening! 2014-11-19 16:50:09 <^7heo> o 2014-11-19 16:50:13 <^7heo> o/ even 2014-11-19 17:57:28 <^7heo> why is there no nmh package for alpine? 2014-11-19 17:57:35 <^7heo> is it because it is GNU? 2014-11-19 18:19:10 ^7heo, nongnu.org/nmh/ ? 2014-11-19 18:19:20 <^7heo> vkrishn: it needs autoshit and all. 2014-11-19 18:19:27 <^7heo> it's freaking complicated. 2014-11-19 18:19:35 <^7heo> I've checked mh rather. 2014-11-19 21:49:26 Yeah, looks like nmh won't build under musl because some of the code diddles the innards of struct FILE (like _cnt and _ptr for instance). 2014-11-19 22:00:25 <^7heo> yeah 2014-11-19 22:00:37 <^7heo> I'm looking forward to build mh 2014-11-19 22:33:35 ^7heo, mh? 2014-11-19 22:33:43 <^7heo> yeah, mh 2014-11-19 22:36:12 ideally it shouldn't need to do that at all 2014-11-19 22:36:42 but if you can't eliminate the need entirely, using the functions from stdio_ext.h would be a way to do the same thing in a less-evil way 2014-11-19 22:37:35 ^7heo, i don't know what mh is 2014-11-19 22:38:11 <^7heo> rand mh 2014-11-19 22:38:23 <^7heo> http://rand-mh.sourceforge.net/ 2014-11-19 23:12:46 interesting 2014-11-19 23:13:00 <^7heo> yeah 2014-11-19 23:13:28 I really don't know much about mail protocols and mail systems though. As a layperson it seems like mail is overly convoluted but don't really know. Thanks. 2014-11-19 23:13:30 <^7heo> offlineimap is good in the sense that it reconstructs a mailbox; but it defeats the purpose of imap, which is to syncronize only on reading 2014-11-19 23:13:57 <^7heo> mail isn't really convoluted. 2014-11-19 23:14:05 <^7heo> it's a very good system 2014-11-19 23:14:16 ehhhhhhh 2014-11-19 23:14:23 <^7heo> but putting trust into mail, along with mail being a very popular thing isn't easy. 2014-11-19 23:14:35 <^7heo> and it brings things like google dominating the freaking markte 2014-11-19 23:14:38 taking in account of security mail is god awful 2014-11-19 23:14:39 <^7heo> market 2014-11-19 23:14:46 <^7heo> and dictating their own rules 2014-11-19 23:14:49 <^7heo> be them good or not. 2014-11-19 23:15:10 <^7heo> absolutely not. 2014-11-19 23:15:18 <^7heo> over SSL with certificate pinning it's not too bad. 2014-11-19 23:15:28 <^7heo> especially if you encrypt your messages in your MUA 2014-11-19 23:15:41 ^7heo, absolutely not w/ respect to what 2014-11-19 23:15:46 <^7heo> ? 2014-11-19 23:15:53 <^7heo> what do you mean? 2014-11-19 23:15:58 <^7heo> in that sense, http sucks too. 2014-11-19 23:15:59 you said absolutely not 2014-11-19 23:16:00 <^7heo> ftp too 2014-11-19 23:16:08 <^7heo> yeah but mail isn't different than the rest 2014-11-19 23:16:12 yah both of those protocols is terrible 2014-11-19 23:16:15 *are 2014-11-19 23:16:21 well http is isn't terrible 2014-11-19 23:16:31 but needs a good kick in the ass 2014-11-19 23:16:33 <^7heo> all protocols are terrible security wise if you don't bring in encryption/authentication. 2014-11-19 23:16:45 sorta 2014-11-19 23:17:15 <^7heo> name a protocol that is good, security wise, while not encompassing encryption and/or authentication 2014-11-19 23:17:16 there are some protocols that make it deliberately easier to work with security mechanisms compared to others 2014-11-19 23:17:42 <^7heo> easier? 2014-11-19 23:17:43 <^7heo> such as? 2014-11-19 23:17:50 <^7heo> I'm not criticizing, I'm just being curious 2014-11-19 23:18:02 yah no offense taken, i enjoy this topic :) 2014-11-19 23:18:24 so despite the issues related around TLS 2014-11-19 23:18:36 <^7heo> so, what protocols make it easier to work with security mechanisms? 2014-11-19 23:18:38 <^7heo> and how? 2014-11-19 23:18:45 the goal feature of segmenting a secure channel 2014-11-19 23:18:53 from the application layer data 2014-11-19 23:18:54 is good 2014-11-19 23:19:09 <^7heo> wait, what? 2014-11-19 23:19:13 were for example snmp 2014-11-19 23:19:18 tries to roll its own encryption 2014-11-19 23:19:28 and just fucks with any notion of logic 2014-11-19 23:19:29 <^7heo> using a custom encryption isn't a good solution. 2014-11-19 23:19:38 <^7heo> you risk to introduce you own flaws. 2014-11-19 23:19:43 i didn't say custom 2014-11-19 23:19:54 i said logical abstraction 2014-11-19 23:19:58 *i meant 2014-11-19 23:19:59 <^7heo> also 2014-11-19 23:20:04 <^7heo> snmp isn't too good. 2014-11-19 23:20:09 oh god 2014-11-19 23:20:15 snmp is horrendous 2014-11-19 23:20:41 i was mentioning snmp on how they didn't use something like TLS or SSL at the time 2014-11-19 23:20:56 to keep the secure channel segmented from it's own application level goals 2014-11-19 23:21:09 or service how ever you want to label your network stack 2014-11-19 23:21:20 <^7heo> snmp v3 'cause snmp v2 or v1 don't even have encryption 2014-11-19 23:21:23 <^7heo> and auth is a joke 2014-11-19 23:21:27 yup 2014-11-19 23:21:42 if i remember correctly v2 2014-11-19 23:21:49 has groups or something 2014-11-19 23:22:02 that is equivalent to telnet passwords 2014-11-19 23:22:20 <^7heo> you mean the private and public community strings? 2014-11-19 23:22:28 <^7heo> that are sent in clear? 2014-11-19 23:22:32 yah 2014-11-19 23:22:36 <^7heo> and that have freaking stupid defaults? 2014-11-19 23:22:43 yah didn't say it is good 2014-11-19 23:22:43 <^7heo> like "private" and "public"? 2014-11-19 23:22:47 yah 2014-11-19 23:22:53 <^7heo> :D 2014-11-19 23:23:05 i did a project related to SNMP a few years ago 2014-11-19 23:23:17 my brain *barfed* 2014-11-19 23:23:29 reading the spec and looking at implementations 2014-11-19 23:23:43 <^7heo> yeah 2014-11-19 23:23:53 <^7heo> and configuring SNMP v3 is a freaking joke 2014-11-19 23:25:21 but yah so issues with mail it didn't really account for encryption, spam, distributed design, anonymity, authenticity 2014-11-19 23:25:22 etc. 2014-11-19 23:25:52 <^7heo> well 2014-11-19 23:25:53 which is understandable for when it was originally created 2014-11-19 23:25:56 <^7heo> mail is a very simple thing 2014-11-19 23:26:01 <^7heo> like ftp 2014-11-19 23:26:07 <^7heo> or http 1.0 2014-11-19 23:26:12 yah 2014-11-19 23:26:13 <^7heo> it can be improved. 2014-11-19 23:26:25 sure but i would argue it would be better to rebase 2014-11-19 23:26:42 than make a Frankenstein monster/ chimera 2014-11-19 23:26:47 <^7heo> but instead of that, people don't want to change the protocol, so they change the server side, adding lots and lots of features. 2014-11-19 23:27:08 yah never understood the fear with changing protocols were 2014-11-19 23:27:12 <^7heo> improving the protocol would be definitely better than doing stupid magic in the server side like google does. 2014-11-19 23:27:19 yarp 2014-11-19 23:27:38 <^7heo> like, I end up in spam on google half of the time 2014-11-19 23:27:40 i am waiting for darkmail to release their spec and implementation 2014-11-19 23:27:41 <^7heo> for no valid reason 2014-11-19 23:27:52 <^7heo> and google is SO dumbed down 2014-11-19 23:27:53 http://darkmail.info/ 2014-11-19 23:28:00 <^7heo> like, regex don't work. ever. no where. 2014-11-19 23:28:20 ^7heo, well I wouldn't assume for no valid reason 2014-11-19 23:28:24 but still sucks 2014-11-19 23:28:29 http://darkmail.info/ 2014-11-19 23:28:43 <^7heo> well, because my mail provider fails to comply with google rules 2014-11-19 23:29:02 <^7heo> which is kinda understandable: why should Google dictate the rules? 2014-11-19 23:29:19 well it isn't really dictating rules 2014-11-19 23:29:23 it's complying with implementation 2014-11-19 23:29:28 *their implementation 2014-11-19 23:29:38 which is their cost benefit decision 2014-11-19 23:30:44 <^7heo> myeah 2014-11-19 23:31:18 <^7heo> damn 2014-11-19 23:31:30 <^7heo> synching my email is SO long... 2014-11-19 23:32:09 ^7heo, that said what they should have done is published an altered spec 2014-11-19 23:32:17 at a minimum 2014-11-19 23:32:22 they have the resources 2014-11-19 23:35:34 ^7heo, do you happen to know 2014-11-19 23:35:47 what needs to be verified for a aports testing packing move into main 2014-11-19 23:35:56 *package 2014-11-19 23:36:33 <^7heo> I guess it is moved to main after it has been in testing for long enough and nobody reported bugs 2014-11-19 23:36:51 <^7heo> you should try ncopa, clandmeter or kaniini for that :) 2014-11-19 23:37:00 m'kay thanks 2014-11-19 23:37:09 just hard sometime since timezon difference :P 2014-11-19 23:37:17 <^7heo> sorry I'm not more helpful 2014-11-19 23:37:40 all good any help is appreciated 2014-11-19 23:38:21 <^7heo> systmkor: what is the apk you're interested in? 2014-11-19 23:38:44 not sure if any particular at the moment 2014-11-19 23:39:04 just was curious what the general standard was 2014-11-19 23:39:13 and if any automation was possible 2014-11-19 23:39:17 <^7heo> ah ok 2014-11-19 23:39:26 <^7heo> it probably is 2014-11-19 23:39:46 <^7heo> I'm personally more interested in distributing the building 2014-11-19 23:39:54 <^7heo> and making apk p2p 2014-11-19 23:40:07 ooh sounds like an interesting problem 2014-11-19 23:40:14 <^7heo> yeah 2014-11-19 23:40:17 <^7heo> in a nutshell 2014-11-19 23:40:22 <^7heo> anyone could publish APKs 2014-11-19 23:40:25 but I also would prefer to move more towards a nix package management port 2014-11-19 23:40:28 <^7heo> with a set of patches 2014-11-19 23:40:40 <^7heo> and then 2014-11-19 23:40:47 <^7heo> you could set you own trustlist 2014-11-19 23:40:55 well dht + git 2014-11-19 23:40:58 <^7heo> yeah 2014-11-19 23:41:02 <^7heo> I wasn't thinking git 2014-11-19 23:41:03 probably would git you a decent way their 2014-11-19 23:41:06 *there 2014-11-19 23:41:06 <^7heo> but DHT for sure 2014-11-19 23:41:18 well being revision control agnostic would probably be best 2014-11-19 23:41:31 <^7heo> I'm not talking about revision control atm 2014-11-19 23:41:42 <^7heo> I don't have any problem with that 2014-11-19 23:41:59 <^7heo> but I'd like to get a system that can build objects, cache them, and share them via p2p 2014-11-19 23:42:03 oh so just just bundling a package and sending that file 2014-11-19 23:42:06 <^7heo> so if someone already built the part you need 2014-11-19 23:42:08 <^7heo> you just get it. 2014-11-19 23:42:12 <^7heo> instead of rebuilding 2014-11-19 23:42:18 yah makes sense 2014-11-19 23:42:23 <^7heo> that way you can doanload every object 2014-11-19 23:42:26 i would be down to help 2014-11-19 23:42:28 <^7heo> then link it 2014-11-19 23:42:29 do you like go-lang? 2014-11-19 23:42:34 <^7heo> I plan to do that in C. 2014-11-19 23:42:35 :D 2014-11-19 23:42:39 noooooooo 2014-11-19 23:42:39 <^7heo> but why not. 2014-11-19 23:42:45 <^7heo> plus 2014-11-19 23:42:49 trust me about go 2014-11-19 23:42:50 <^7heo> don't take my word on that. 2014-11-19 23:42:56 <^7heo> It's just an idea. 2014-11-19 23:42:57 if its above kernel & drivers 2014-11-19 23:43:02 go / rust is the way to go 2014-11-19 23:43:11 <^7heo> I don't have the time atm. 2014-11-19 23:43:12 though rust is a few years too early at the moment 2014-11-19 23:43:18 <^7heo> so I'm not going to say "let's do it" 2014-11-19 23:43:24 <^7heo> even if I'm very interested. 2014-11-19 23:43:37 <^7heo> now if you wanna do it, knock yourself :P 2014-11-19 23:43:51 *knock knock* 2014-11-19 23:43:54 <^7heo> :P 2014-11-19 23:44:02 it actually is a related problem to what I am doing at work 2014-11-19 23:44:02 <^7heo> my point is very simple 2014-11-19 23:44:10 <^7heo> gentoo is one of the best systems I've used but 2014-11-19 23:44:18 <^7heo> 1. compilation takes WAY too long 2014-11-19 23:44:21 yup 2014-11-19 23:44:25 <^7heo> even on high end stuff 2014-11-19 23:44:42 <^7heo> 2. it's a pain to setup everything to get your top-notch linux; and even more of a pain to share that. 2014-11-19 23:44:46 <^7heo> so 2014-11-19 23:44:50 ^7heo, what you don't have 64 core 8 GHz system? 2014-11-19 23:44:53 :P 2014-11-19 23:45:00 <^7heo> no I don't. 2014-11-19 23:45:06 <^7heo> only a 8 core 3.8 GHz 2014-11-19 23:45:10 ddammnnn 2014-11-19 23:45:12 <^7heo> yeah 2014-11-19 23:45:25 <^7heo> to solve 1. I plan to build and store, then hash and share. 2014-11-19 23:45:29 ^7heo, the other thing that would be nice is gamifying packagin 2014-11-19 23:45:42 <^7heo> obviously, sharing the hashes, then sharing the content to whoever is interested. 2014-11-19 23:45:54 with gamification on top 2014-11-19 23:46:00 <^7heo> to solve 2. I plan to make ANY system able to pull ANY receipe from ANYone. 2014-11-19 23:46:03 build trust webs 2014-11-19 23:46:06 validate builds 2014-11-19 23:46:10 <^7heo> so I can build my own linux, with whatever I want 2014-11-19 23:46:11 distribute computation 2014-11-19 23:46:14 <^7heo> the day I stop 2014-11-19 23:46:16 distribute file transfer 2014-11-19 23:46:25 <^7heo> people will use other sources. 2014-11-19 23:46:26 and see if we could work in code audit and review 2014-11-19 23:46:27 <^7heo> done. 2014-11-19 23:46:31 <^7heo> yeah 2014-11-19 23:46:38 <^7heo> then people won't do makefiles anymore. 2014-11-19 23:46:44 <^7heo> they will just do source code. 2014-11-19 23:47:02 <^7heo> and they will just wait for whoever to come with the best build method. 2014-11-19 23:47:08 well build systems are still at a minimum language dependent 2014-11-19 23:47:17 <^7heo> cmake? 2014-11-19 23:47:25 <^7heo> m4? 2014-11-19 23:47:29 errrhhhhhhhhhhh 2014-11-19 23:47:34 those kinda work for C 2014-11-19 23:47:36 <^7heo> :D 2014-11-19 23:47:39 <^7heo> yeah 2014-11-19 23:47:43 go for example has it's own build system 2014-11-19 23:47:46 <^7heo> make/bsdmake is great. 2014-11-19 23:47:47 that works relatively 2014-11-19 23:47:53 identical on any platform 2014-11-19 23:47:54 <^7heo> yeah but that sucks imho 2014-11-19 23:48:04 which is awesome 2014-11-19 23:48:05 <^7heo> it's dependent on Go 2014-11-19 23:48:09 <^7heo> which sucks 2014-11-19 23:48:14 well you could build your own 2014-11-19 23:48:16 <^7heo> yeah 2014-11-19 23:48:17 just as CMake did 2014-11-19 23:48:21 <^7heo> that's the point. 2014-11-19 23:48:28 <^7heo> I want to P2P UNIX 2014-11-19 23:48:36 <^7heo> so people stop rewriting the wheel 2014-11-19 23:48:37 I agreee 2014-11-19 23:48:42 <^7heo> even if they don't reinvent it 2014-11-19 23:48:43 but building != package install 2014-11-19 23:48:46 <^7heo> for now, they must rewrite. 2014-11-19 23:48:50 <^7heo> no 2014-11-19 23:49:01 this conversation makes me very happy i run freebsd on most things now 2014-11-19 23:49:02 <^7heo> that's where APK comes 2014-11-19 23:49:09 <^7heo> kaniini: why? 2014-11-19 23:49:25 because i would not want to use some p2p shit 2014-11-19 23:49:32 <^7heo> kaniini: you think that a central authority that controls "ports" is better than a P2P network? 2014-11-19 23:49:44 <^7heo> kaniini: you can do central auth with P2P 2014-11-19 23:49:48 ^7heo kaniini , with a peer to peer design 2014-11-19 23:49:50 <^7heo> auth as in authority 2014-11-19 23:49:54 you can opt in to a central authority 2014-11-19 23:49:58 <^7heo> exactly 2014-11-19 23:50:01 <^7heo> but not the contrary 2014-11-19 23:50:08 <^7heo> I'm just opting for more freedom. 2014-11-19 23:50:15 kaniini, and in some cases like enterprise that is probably best 2014-11-19 23:50:43 <^7heo> if you want to stick with freeBSD.official that system will work as well as your current one. 2014-11-19 23:50:52 <^7heo> but if you want to experiment 2014-11-19 23:50:57 <^7heo> if you want to provide your own stuff 2014-11-19 23:51:07 <^7heo> if you want to use your cousin's packages 2014-11-19 23:51:19 <^7heo> if you want to go for suckless's morpheus statically built stuff 2014-11-19 23:51:20 or have your own private environment with friends 2014-11-19 23:51:21 <^7heo> whatever 2014-11-19 23:51:24 <^7heo> you could. 2014-11-19 23:51:26 <^7heo> at least. 2014-11-19 23:51:28 it would take just configuraiton modification 2014-11-19 23:51:31 <^7heo> yeah 2014-11-19 23:51:31 not a whole new system 2014-11-19 23:51:34 <^7heo> exactly 2014-11-19 23:51:37 <^7heo> that's the WHOLE point. 2014-11-19 23:51:51 <^7heo> I want the concept of Linux distribution to die, basically. 2014-11-19 23:52:12 <^7heo> 'cause it's just, all the time, about concentrating the power around people who don't know the end users 2014-11-19 23:52:13 ^7heo, totally behind that 2014-11-19 23:52:22 <^7heo> who become egotistic maniacs 2014-11-19 23:52:24 <^7heo> and who do shit 2014-11-19 23:52:31 well that definitely is case dependent 2014-11-19 23:52:36 <^7heo> yeah. 2014-11-19 23:52:37 but the bigger issue I would say 2014-11-19 23:52:40 <^7heo> of course. 2014-11-19 23:52:43 is that it silo's work 2014-11-19 23:52:53 rather all pushing back to mainline so to speak 2014-11-19 23:52:55 <^7heo> but the most important point is: it can work in the way it works now. 2014-11-19 23:53:12 <^7heo> and hope to not add too much overhead if it has to work in a centralized way. 2014-11-19 23:53:35 <^7heo> 'cause in that case, you don't need DHT or anything. 2014-11-19 23:53:42 <^7heo> but whatever. 2014-11-19 23:53:44 <^7heo> that's just a dream. 2014-11-19 23:53:48 <^7heo> for now, I have to code my stuff 2014-11-19 23:53:51 <^7heo> make it work 2014-11-19 23:53:54 <^7heo> and then I'll consider that. 2014-11-19 23:54:14 also what would be cool is adding in distributed message queues 2014-11-19 23:54:26 <^7heo> what do you mean? 2014-11-19 23:54:30 so example: susbsribe to the official Alpine build 2014-11-19 23:54:40 and you can automatically build code in a LXC container 2014-11-19 23:54:42 for the community 2014-11-19 23:54:48 <^7heo> that's basically docker. 2014-11-19 23:54:51 without having to know how to do any manual work 2014-11-19 23:55:04 well you use LXC to reduce security issues 2014-11-19 23:55:15 but the actuall packaging would be the DHT build system 2014-11-19 23:55:20 inside the linux container 2014-11-19 23:55:25 not sending the linux container itself 2014-11-19 23:55:27 like docker 2014-11-19 23:55:28 <^7heo> my goal is just to transform "trust to unknown authorities" into "trust to people you know physically", when you want that. 2014-11-19 23:55:42 i agree for that 2014-11-19 23:55:52 just adding new by-product benefits 2014-11-19 23:55:54 <^7heo> 'cause, with all due respect, and in spite of the quality of the distribution; why should I trust ncopa, clandmeter and kaniini ? 2014-11-19 23:55:59 <^7heo> because they do a good work now? 2014-11-19 23:56:01 the Linux/UNIX community 2014-11-19 23:56:03 to get from that 2014-11-19 23:56:20 <^7heo> what guarantees me that they won't go berserk and add their own CA to MITM me in 1 year? 2014-11-19 23:56:35 <^7heo> and that's just an example. 2014-11-19 23:56:42 while technically true 2014-11-19 23:56:49 <^7heo> now, if I pull my code/binaries from people I trust 2014-11-19 23:56:49 there isn't infinite resources 2014-11-19 23:56:58 ^7heo, because the source and binaries are widely distributed, the source is in version control, and if they intentionally introduced malicious code, there would be an audit trail pointing back to them when it was caught 2014-11-19 23:56:59 so the majority of people have to trust someone at some point 2014-11-19 23:57:00 <^7heo> I'll sleep much better. 2014-11-19 23:57:14 <^7heo> dalias: yeah, right. 2014-11-19 23:57:23 <^7heo> like heartbleed was spotted right away. 2014-11-19 23:57:31 <^7heo> not after 2 years. 2014-11-19 23:57:37 ^7heo, well that is were gamifying building 2014-11-19 23:57:41 I think is important 2014-11-19 23:57:42 it's unclear whether heartbleed was intentional 2014-11-19 23:57:47 but it's still quite different 2014-11-19 23:57:48 <^7heo> imho it was. 2014-11-19 23:57:48 were most points is awarded for code audit 2014-11-19 23:57:53 <^7heo> and yes it's quite different. 2014-11-19 23:57:55 <^7heo> but as I said 2014-11-19 23:57:59 a distro does not make major architectural changes to their packages 2014-11-19 23:58:00 <^7heo> it's just an example. 2014-11-19 23:58:08 the patches are small 2014-11-19 23:58:22 and easy to audit for blame 2014-11-19 23:58:22 <^7heo> I can tell you that if you inject your own CA into the cert list... 2014-11-19 23:58:33 <^7heo> less than a percent of users will notice anything 2014-11-19 23:58:35 <^7heo> even on alpine. 2014-11-19 23:58:48 the question isn't whether they'll notice immediately 2014-11-19 23:58:48 <^7heo> and I'm not even considering noobunto or arschlinux. 2014-11-19 23:59:04 the question is whether there's an audit trail pointing to the guilty person when someone does notice 2014-11-19 23:59:12 <^7heo> yeah 2014-11-19 23:59:20 and the answer is yes 2014-11-19 23:59:22 <^7heo> like I totally care who is to blame. 2014-11-19 23:59:32 it disincentivizes such malice 2014-11-19 23:59:35 <^7heo> once my data is compromised, I don't care about shit. 2014-11-19 23:59:47 <^7heo> to some extent. 2014-11-20 00:00:02 <^7heo> I'm sorry, this is the same as saying: changing the port number of ssh is a security protection. 2014-11-20 00:00:09 no it's not 2014-11-20 00:00:13 <^7heo> yeah it is. 2014-11-20 00:00:19 <^7heo> if people want to compromize you 2014-11-20 00:00:23 agreeing with dalias it's not quite the same 2014-11-20 00:00:24 <^7heo> they can gain trust slowly 2014-11-20 00:00:37 <^7heo> they can deceive you when they NEED it. 2014-11-20 00:00:38 ^7heo, yes the audit trail is nice 2014-11-20 00:00:41 <^7heo> and then take the blame. 2014-11-20 00:00:44 <^7heo> the deed is done. 2014-11-20 00:00:44 however it doesn't do preventative 2014-11-20 00:00:54 ^7heo, care to cite an example? 2014-11-20 00:00:56 it does provide a post analysis though 2014-11-20 00:01:00 <^7heo> dalias: at 1 AM? no. 2014-11-20 00:01:02 <^7heo> I don't care. 2014-11-20 00:01:06 <^7heo> I'm not crawling the web for that. 2014-11-20 00:01:11 <^7heo> I made my point clear. 2014-11-20 00:01:49 agreed 2014-11-20 00:01:52 http://xkcd.com/538/ 2014-11-20 00:01:55 <^7heo> and furthermore I don't need an example to make my point valid. 2014-11-20 00:01:57 okay so ^7heo what were you thinking 2014-11-20 00:02:05 other than DHT 2014-11-20 00:02:08 <^7heo> and xkcd is full of fallacy and shit 2014-11-20 00:02:11 <^7heo> funny ones, yes. 2014-11-20 00:02:18 <^7heo> but fallacy and shit nevertheless. 2014-11-20 00:02:21 you do need an example because it's a nonsensical hypothetical example 2014-11-20 00:02:40 <^7heo> systmkor: not really defined anything yet. 2014-11-20 00:02:46 <^7heo> I just got the idea in the corner of my head. 2014-11-20 00:02:52 <^7heo> I want a distributed process. 2014-11-20 00:03:01 m'kay well I'll think about it 2014-11-20 00:03:02 <^7heo> 'cause right now it's what's killing all good linux distros 2014-11-20 00:03:07 <^7heo> building and maintainning. 2014-11-20 00:03:14 <^7heo> we need to be able to distribute that when needed. 2014-11-20 00:03:15 <^7heo> like 2014-11-20 00:03:20 <^7heo> take the kernel from 2014-11-20 00:03:25 <^7heo> the libc from 2014-11-20 00:03:26 <^7heo> etc. 2014-11-20 00:04:26 <^7heo> dalias: yeah, like nothing ever goes wrong with trusting people you don't know. 2014-11-20 00:04:31 <^7heo> anyway 2014-11-20 00:04:35 <^7heo> imap synchronized. 2014-11-20 00:04:37 <^7heo> and I need rest. 2014-11-20 00:04:54 nobody with the determination to attack you is going to do it via an inefficient, unreliable, super-slow method that also affects lots of non-targets, each of whom poses a risk of the attack being discovered 2014-11-20 00:05:09 they're going to use a much more direct, targeted attack 2014-11-20 00:05:28 dalias, well it depends if it is intended to be a target attack 2014-11-20 00:05:30 but I agree 2014-11-20 00:05:35 the quantified risk 2014-11-20 00:05:38 is very low 2014-11-20 00:06:13 <^7heo> dalias: nobody. just nobody. I can tell some sentence that contains "nobody" and is true at the same time, however: 2014-11-20 00:06:18 <^7heo> behold: 2014-11-20 00:06:45 <^7heo> nobody who has at least the pretention to be serious about security will EVER use "nobody" in any case when it comes to a potential case. 2014-11-20 00:07:15 <^7heo> plus, the CA list is fucking huge and complex. 2014-11-20 00:07:27 <^7heo> adding and entry somewhere can lie undetected for YEARS 2014-11-20 00:07:45 <^7heo> and the person can have the time to hide and erase traces. 2014-11-20 00:07:58 <^7heo> before it even remotely comes dangerous, because it's not used. 2014-11-20 00:08:00 <^7heo> yet. 2014-11-20 00:08:27 <^7heo> that said 2014-11-20 00:08:31 <^7heo> sleeping time, really. 2014-11-20 00:08:35 g'night 2014-11-20 00:08:36 <^7heo> even the computer wanna go to sleep 2014-11-20 00:08:36 ttyl 2014-11-20 00:08:40 <^7heo> o/ 2014-11-20 00:18:26 ^7heo, when packages are maintained in git, no 2014-11-20 00:18:35 there's no hiding or erasing traces 2014-11-20 00:18:48 and what you'll see in the distro's package files is the diff from the upstream CA list 2014-11-20 00:18:52 not the whole list 2014-11-20 00:19:00 and thus you'll immediately see the malicious additions 2014-11-20 00:26:03 dalias, hey I am still running into timeval struct issues 2014-11-20 00:26:04 one sec 2014-11-20 00:26:54 dalias, I am trying to compile telehash-c on Alpine 2014-11-20 00:27:07 and that is were I am getting the timeval not defined issue 2014-11-20 00:27:18 even when I include bits/alltypes.h 2014-11-20 00:27:21 https://github.com/telehash/telehash-c 2014-11-20 00:27:31 dalias, https://github.com/telehash/telehash-c/issues/24 2014-11-20 00:29:27 you don't include bits/alltypes.h :) 2014-11-20 00:29:35 that's part of the implementation of other headers 2014-11-20 00:29:38 you include 2014-11-20 00:29:38 ah 2014-11-20 00:29:43 hrmm one sec 2014-11-20 00:29:49 oops timeval 2014-11-20 00:29:52 misread that as timespec 2014-11-20 00:30:08 timeval is or or a number of other "old unixy" headers 2014-11-20 00:32:17 up i think it works now 2014-11-20 00:32:18 thanks 2014-11-20 00:33:43 looks like the author fixed it this morning 2014-11-20 00:33:50 he put in 2014-11-20 00:34:15 dalias, has anyone made a tool to automatically tell you the mapping from glib #include to musl 2014-11-20 00:48:06 it's the same on glibc, but in the default feature profile several glibc headers randomly pull in each other for no good reason 2014-11-20 01:30:19 dalias, well I guess more of some script so developers for a project 2014-11-20 01:30:29 can just make it stupid simple to make it work with MUSL 2014-11-20 07:01:39 systmkor: the tool is awk or grep and http://port70.net/~nsz/c/posix/reserved.txt 2014-11-20 07:06:38 (which shows sys/select.h should be used with _POSIX_C_SOURCE and no XSI extension) 2014-11-20 07:10:17 nsz, thanks 2014-11-20 07:54:58 just upgraded to xen 4.3.3 in AL 2.7.9 and rebooted yesterday and this morning all my VMs are "hung" :( 2014-11-20 07:55:24 oh well, at least i'll be busy today :) 2014-11-20 10:12:03 Hello, how can one 'reload' Busybox's 'init'? Similar to 'telinit q' ? 2014-11-20 16:21:41 My /var/log/messages is rotating every 1 min 2014-11-20 16:21:50 I dont remember where to change this? 2014-11-20 16:22:10 every 200M 2014-11-20 16:23:12 <^7heo> ?? 2014-11-20 16:24:30 alacerda: you can modify SYSLOGD_OPTS in /etc/conf.d/syslog 2014-11-20 16:24:58 pnutzh4x0r, thx 2014-11-20 16:24:59 if you are using the default syslogd from busybox 2014-11-20 16:25:13 yes that is what i was looking for 2014-11-21 09:04:08 morning 2014-11-21 09:14:53 <^7heo> moin 2014-11-21 13:11:34 To the people dealing with the website UI revamp, I made a thread in order to get graphical designers' opinions. There's been some constructive criticism so far. 2014-11-21 13:11:35 http://boards.4chan.org/gd/thread/186089 2014-11-21 13:27:48 <^7heo> diwerth: nice. 2014-11-21 13:39:57 I like the proposed logo 2014-11-21 13:40:10 Simple and stylish 2014-11-21 13:40:39 Well. The core idea is simple. I'd make it as simple as possible 2014-11-21 13:57:34 at this point i am more interested to shoot the enginners (and designers) and just ship it :) 2014-11-21 13:57:47 http://wwwtest.alpinelinux.org 2014-11-21 13:57:53 mostly usable 2014-11-21 13:59:52 <^7heo> ncopa: it's strange to have a different height in the menu "current" item depending on what page you are on. 2014-11-21 14:01:41 I think I prefer the current website, but it may just be a matter of getting used to the new one 2014-11-21 14:02:06 ncopa, the logo on the test page is nice and exactly what I was talking about! 2014-11-21 14:02:09 :3 2014-11-21 14:13:37 our current website is much better! 2014-11-21 14:13:46 ;-) 2014-11-21 14:29:20 on the wwwtest, i have a utf8 char that is not displayed correctly. can there be some fallback for peops not having the font? 2014-11-21 14:29:33 the top left utf char on the page 2014-11-21 14:30:51 otherwise my opionion is irrelevant. usually if i like something it is a flop, if i do not like something it will be a hit. i feel neutral towards this design. ;) 2014-11-21 14:46:29 can anyone confirm that its not possible to login to alpine if you use 1 char long usernames (additionally also 1 char long hostname is used) 2014-11-21 14:46:39 and i'm unable to login as this one-char user 2014-11-21 14:46:57 also i get strange warnings when changing the password of this user as root 2014-11-21 14:47:19 Hi 2014-11-21 14:47:27 I want to install unstable firefox apk 2014-11-21 14:47:38 in stable Alpine Linux 2014-11-21 14:47:45 is it possible? 2014-11-21 14:48:06 how? 2014-11-21 14:50:06 chatr: http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2014-11-21 14:51:41 thanks 2014-11-21 14:51:51 what if I've already downloaded the apk? 2014-11-21 14:52:06 How can I install it using apk command? 2014-11-21 14:52:34 Is it possible? 2014-11-21 15:01:10 If I install firefox@edge over firefox 2014-11-21 15:01:15 What happens? 2014-11-21 15:01:20 A conflict? 2014-11-21 15:02:11 hmm 2014-11-21 15:02:14 it's upgrading 2014-11-21 16:49:28 I've used repository pinning 2014-11-21 16:49:29 http://wiki.alpinelinux.org/wiki/Alpine_Linux_package_management#Repository_pinning 2014-11-21 16:49:37 to install latest firefox 2014-11-21 16:50:07 sudo apk add firefox@testing icu-libs@edge 2014-11-21 16:50:36 because only sudo apk add firefox@testing didn't work and it complained about lack of latest icu-libs 2014-11-21 16:50:48 installation worked fine 2014-11-21 16:50:56 but when I try to run firefox 2014-11-21 16:51:00 it doesn't work 2014-11-21 16:51:11 XPCOMGlueLoad error for file /usr/lib/firefox-33.1.1/libmozalloc.so: 2014-11-21 16:51:16 Error relocating /usr/lib/firefox-33.1.1/libmozalloc.so: malloc_usable_size: symbol not found 2014-11-21 16:51:21 Couldn't load XPCOM. 2014-11-21 16:53:28 So, what's the problem with libmozalloc.so? 2014-11-21 16:56:01 you probably not using latest musl 2014-11-21 16:56:30 upgrade musl package and it should work 2014-11-21 16:56:42 But does it break other packages? 2014-11-21 16:56:53 alpine seems to have some problems with missing dependencies like this 2014-11-21 16:56:57 it shouldn't 2014-11-21 16:57:08 I'm afraid everything is compiled with older version of musl 2014-11-21 16:57:14 that's fine 2014-11-21 16:57:30 I'm trying this 2014-11-21 16:57:31 compiled-with-older-version should always work with newer-version 2014-11-21 16:57:33 sudo apk add musl@edge 2014-11-21 16:57:37 it's the other way around that might not work 2014-11-21 16:58:09 Isn't that old DLL hell people encountered in Windows? 2014-11-21 16:58:27 I think they've worked around this problem 2014-11-21 16:58:38 by allowing installation of different versions of the same library 2014-11-21 16:58:38 no 2014-11-21 16:58:41 Isn't it? 2014-11-21 16:58:59 Could you please explain? 2014-11-21 16:59:09 dll hell is when you have different versions of a library with incompatible interfaces with the same name 2014-11-21 16:59:18 and thus some programs NEED the old one and others NEED the new one 2014-11-21 16:59:37 it's a result of bad library design and maintenance practices 2014-11-21 17:00:39 But, this can also happen in linux or in general FOSS libraries, isn't it? 2014-11-21 17:01:11 hmm 2014-11-21 17:01:14 Now it runs 2014-11-21 17:01:16 but it says 2014-11-21 17:01:18 The application has been updated, but your version of SQLite is too old and the application cannot run. 2014-11-21 17:01:24 I'm upgrading SQLite 2014-11-21 17:03:43 What should I upgrade for SQLite? 2014-11-21 17:04:57 i'm not sure 2014-11-21 17:05:02 there is sqlite-libs, sqlite and also sqlite-dev, I upgraded them all, but I still get that message from firefox, and also running sqlite3 command I get SQLite header and source version mismatch 2014-11-21 17:05:24 How can I check which version of a package is installed? 2014-11-21 17:08:09 apk version? 2014-11-21 17:08:34 it gives blank results 2014-11-21 17:08:39 Installed: Available: 2014-11-21 17:09:27 sorry i'm not an expert on using apk 2014-11-21 17:09:47 i just use edge and things mostly work 2014-11-21 17:11:46 I chose stable and now I think I should switch to edge 2014-11-21 17:11:50 to avoid problems like this 2014-11-21 17:14:29 Ok! 2014-11-21 17:14:34 I upgraded sqlite-libs 2014-11-21 17:15:37 and firefox worked fine 2014-11-21 17:15:50 I had to run it in safe mode 2014-11-21 17:16:05 because it seems that it does not work with some old plugins/extensions 2014-11-21 17:18:14 I upgraded firefox 2014-11-21 17:18:32 but It seems that is not a good practice 2014-11-21 17:18:49 because there is a chance to break packages 2014-11-21 17:46:40 I'm thinking about Adobe Flash and other propertiary software 2014-11-21 17:46:44 available for Linux 2014-11-21 17:47:01 Is there a chance of using Adobe Flash on Alpine Linux? 2014-11-21 17:47:36 Or we should wait until Musl provides binary compatibility with GLibc? 2014-11-21 17:49:00 sure, in a glibc compartment 2014-11-21 17:49:11 haha. good luck waiting for that. 2014-11-21 17:49:21 i think that's no goal 2014-11-21 17:50:29 even steve got rid of flash, why would we stick to that? 2014-11-21 17:50:44 chatr: wont happen. the abi is half the bloat of glibc. 2014-11-21 17:51:30 I'm puzzled by the LOC of glibc compared to musl 2014-11-21 17:52:02 https://www.openhub.net/p/glibc 2014-11-21 17:52:32 Glibc > 2.4 Million lines of code 2014-11-21 17:52:51 Musl > 84 thousand lines of code 2014-11-21 17:53:30 Do they provide comparable functionality?! 2014-11-21 17:55:14 they do. except the gnuisms 2014-11-21 17:56:04 and the drepperisms 2014-11-21 17:57:03 i mean they must provide comparable functionality if systems run on both 2014-11-21 17:57:40 Glibc codebase is 28 times bigger 2014-11-21 17:58:00 and that is really questionable 2014-11-21 17:59:41 you forget glibc has a huge age-advantage when it comes to bitrot, feature creep, etc 2014-11-21 18:07:15 :-) 2014-11-21 18:10:34 The reverse path seems to be more promising. 2014-11-21 18:10:57 Canonical developed a compatibility layer 2014-11-21 18:11:03 called libhybris 2014-11-21 18:11:27 that translates Bionic-libc calls 2014-11-21 18:11:32 to glibc calls 2014-11-21 18:11:38 ugh 2014-11-21 18:11:47 It seems that canonical wants to run android apps on their devices 2014-11-21 18:11:56 Jolla did that. 2014-11-21 18:12:10 well, glibc is still golden compared to bionic 2014-11-21 18:12:32 "We get it google, android IS NOT UNIX, just because you say so" 2014-11-21 18:12:38 Jolla claims that they can run Android apps on their device 2014-11-21 18:12:45 except the kernel. and everything. just not their broken userspace :/ 2014-11-21 18:13:06 Yeah. 2014-11-21 18:13:09 I agree. 2014-11-21 18:13:10 :-) 2014-11-21 18:13:23 chatr: its not that difficult to run android locally 2014-11-21 18:13:29 How? 2014-11-21 18:13:31 even with qemu with syscall emulation 2014-11-21 18:13:38 yeah 2014-11-21 18:13:43 But that's really annoying 2014-11-21 18:13:53 depends what you need to run 2014-11-21 18:14:04 I think it should be possible now 2014-11-21 18:14:14 that ART compiles apps into binaries 2014-11-21 18:14:18 and run the binary 2014-11-21 18:14:23 the whole dalvik ordeal is probably a bit more trickier, than just separate .so you need to gut from out of there :) 2014-11-21 18:14:23 We have libhybris 2014-11-21 18:14:41 hu? 2014-11-21 18:14:42 But as I understood 2014-11-21 18:14:51 either you emulate whole dalvik 2014-11-21 18:14:59 or no android apps as you know em 2014-11-21 18:15:01 theres no way around this 2014-11-21 18:15:02 I'm talking about ART 2014-11-21 18:15:24 no idea what that is 2014-11-21 18:15:26 it is set as default, replacing Dalvik from Android 5.0 2014-11-21 18:15:32 It compiles Java bytecode 2014-11-21 18:15:34 into binary 2014-11-21 18:15:37 ah 2014-11-21 18:15:38 neat 2014-11-21 18:15:39 and then runs it 2014-11-21 18:15:42 but thats not the problem 2014-11-21 18:15:49 But unfortunately 2014-11-21 18:15:51 problem is whole java runtime and UI you need 2014-11-21 18:16:00 yeah 2014-11-21 18:16:05 but with libhybris 2014-11-21 18:16:20 if that can be emulated, thats fairly impressive 2014-11-21 18:16:23 the translation into Waland calls is done 2014-11-21 18:16:23 assuming its cleanroom 2014-11-21 18:16:29 and not just qemu for whole machine 2014-11-21 18:16:35 chatr, supposedly flash works but i haven't confirmed this. you might have to do some minor hacking on the .so file 2014-11-21 18:16:35 The stupid thing here 2014-11-21 18:17:08 is that google ART creates its own binary format 2014-11-21 18:17:23 :) 2014-11-21 18:17:24 chatr: binfmt, or just loader? 2014-11-21 18:17:26 and not normal Linux binary 2014-11-21 18:17:32 chatr: if its just loader, theres no problem there 2014-11-21 18:17:37 its no different from dalvik, really 2014-11-21 18:17:57 as long theres entrypoint you can throw at qemu, what happens after that is fairly accurate 2014-11-21 18:18:07 as long you can fake the ui IO that is 2014-11-21 18:18:39 But with qemu, you should have a whole image of an Android device 2014-11-21 18:18:45 to open a simple app 2014-11-21 18:19:00 and that's really a pain 2014-11-21 18:19:07 what? no 2014-11-21 18:19:14 of course we're talking user emulation :) 2014-11-21 18:19:22 I don't know about that 2014-11-21 18:19:26 User emulation? 2014-11-21 18:19:37 yes, qemu can take foreign elf binary 2014-11-21 18:19:39 and just run it 2014-11-21 18:19:44 translating syscalls, using the host kernel directly. 2014-11-21 18:19:57 it works for all sorts of anroid garbage 2014-11-21 18:20:09 :-)) 2014-11-21 18:20:12 as long one can fake the interfaces it expects 2014-11-21 18:20:34 That's great. :-) 2014-11-21 18:20:51 well, thats necessary only for arm binaries obviously 2014-11-21 18:21:00 for x86 android binaries ... well, run it directly :) 2014-11-21 18:21:07 in chroot or lxc 2014-11-21 18:21:53 So, I should be able to run the compiled apps, on the fly, if I have the appropriate libs in a chroot? 2014-11-21 18:22:30 yes, android linux kernel is still a linux 2014-11-21 18:22:34 and host still is also a linux 2014-11-21 18:23:09 however this sort of thing fails horribly out of the box, as it expects bunch of io pipes laying around to talk to ui and daemon processes 2014-11-21 18:23:41 you either have to run those too (possible, but difficult) or fake it (cleanroom host os, probably preferable) 2014-11-21 18:24:50 So, the main problem is not the binary format, but the runtime dependencies. Am I right? 2014-11-21 18:25:01 yup 2014-11-21 18:25:18 Google itself now compiles these binaries into Javascript 2014-11-21 18:25:21 the runtime format is the lowest level part and fairly easy one. google didnt muck with the low level details too much. 2014-11-21 18:25:26 and run them on Google Chrome 2014-11-21 18:25:44 well, i thought it was as pnacl in chromeos? 2014-11-21 18:25:54 yeah 2014-11-21 18:25:58 it runs on NaCl 2014-11-21 18:26:16 Some apps don't run 2014-11-21 18:26:21 well, that could be possibly good starting point for reuse 2014-11-21 18:26:29 But most of them do 2014-11-21 18:26:42 since nacl is neatly isolated already ... 2014-11-21 18:27:00 might just run the beast standalone without chromeos and thats it :) 2014-11-21 18:27:20 Yeah 2014-11-21 18:27:26 That would be great! 2014-11-21 18:27:47 didnt take a look how exactly it works 2014-11-21 18:27:48 If I compile Alpine Linux with damn Bionic libc 2014-11-21 18:27:58 they package the whole app as nacl blob? 2014-11-21 18:28:00 I should be able to run the apps 2014-11-21 18:28:32 why would anyone want to voluntarily use bionic? 2014-11-21 18:28:54 that stuff is sort of bare metal maybe-ansic toolkit 2014-11-21 18:29:10 definitely not a standard system library, with 3/4 the stuff missing... 2014-11-21 18:30:08 read this 2014-11-21 18:30:11 http://arstechnica.com/gadgets/2012/02/chrome-finally-brings-modern-web-standards-to-android/ 2014-11-21 18:30:57 There is a lot of incompatibilities and of course compatibility layers in Android 2014-11-21 18:31:06 I was wondering how does Android apps 2014-11-21 18:31:09 compiled for ARM 2014-11-21 18:31:17 work on Intel Atom based tablets 2014-11-21 18:31:46 and I understood that there's a compatibility layer that translates ARM instructions into Intel instructions! 2014-11-21 18:31:57 So, a lot of your CPU power 2014-11-21 18:32:10 is wasted on crapware that these guys create 2014-11-21 18:32:25 to be able to run existing apps 2014-11-21 18:32:36 :-( 2014-11-21 18:38:58 I can't find the Github page for the project 2014-11-21 18:39:08 which helps run Android apps on Chrome 2014-11-21 18:39:38 Ok 2014-11-21 18:39:44 I found it 2014-11-21 18:39:44 ARChon 2014-11-21 18:40:11 I could not find its Github page 2014-11-21 18:40:20 because it is on Bitbucket! 2014-11-21 18:41:02 https://bitbucket.org/vladikoff/archon/ 2014-11-21 18:41:02 This is some explanation 2014-11-21 18:41:11 http://lifehacker.com/how-to-run-android-apps-inside-chrome-on-any-desktop-op-1637564101 2014-11-21 18:41:18 it depends on The App Runtime for Chrome (or ARC) 2014-11-21 18:55:22 I don't know if it's appropriate to continue talking about this here 2014-11-21 18:56:23 But one last note 2014-11-21 18:56:32 take a look at here 2014-11-21 18:56:33 https://bitbucket.org/vladikoff/archon/src/4369d5a9faba54d963be5bdd06b7c857b7eab052/_platform_specific/nacl_x86_64/?at=master 2014-11-21 18:56:49 It seems that these libraries 2014-11-21 18:56:59 are all that an Android app needs 2014-11-21 18:57:04 most apps 2014-11-21 18:57:13 need to be able to run correctly 2014-11-21 19:26:51 katu: Where to talk more about running Android apps on Linux? 2014-11-21 19:27:32 #alpine-devel ? 2014-11-21 19:28:52 i think the devel channel is for actually working on development 2014-11-21 19:31:21 maybe android channels 2014-11-21 19:32:16 I'm glad if we can continue talking about it here 2014-11-21 19:32:17 :-) 2014-11-21 19:40:38 ok 2014-11-21 19:42:59 So, I wanna see what's the obstacle here? 2014-11-21 19:43:08 Considering that we are working on an ARM machnie 2014-11-21 19:43:11 rer 2014-11-21 19:44:09 and using Bionic libc 2014-11-21 19:44:25 being able to run ART runtime 2014-11-21 19:44:52 we can convert the Java app into some sort of ELF binary 2014-11-21 19:46:41 we have most (if not all) needed libraries from ARC (App Runtime for Chrome) or ARChon 2014-11-21 19:47:03 So, what's the problem here? :) 2014-11-21 19:51:48 This is for ARM 2014-11-21 19:51:51 https://bitbucket.org/vladikoff/archon/src/b8907528b35de47d9175663de151e1f9258cf1f2?at=arm_1.2 2014-11-21 19:52:04 chatr: dont talk, just try it :) 2014-11-21 19:52:19 I should find some ARM platform first 2014-11-21 19:52:25 maybe raspberry pi 2014-11-21 19:52:35 as for bionic - thats what the apps are linked against. you dont, repeat dont, need host os linked with bionic. 2014-11-21 19:52:37 recompile Alpine Linux using Bionic 2014-11-21 19:53:09 Do you think recompiling Alpine with Bionic is possible with little effort? 2014-11-21 19:53:09 ... 2014-11-21 19:53:18 definitely not 2014-11-21 19:53:24 bionic is utterly broken and unusable 2014-11-21 19:53:38 Do you alter apps much to compile them with Musl instead of Glibc? 2014-11-21 19:53:44 all it's made to do is host a jvm 2014-11-21 19:54:10 chatr: do you think we should compile alpine linux with msvcrt.dll, because that sounds like good idea when using wine? 2014-11-21 19:54:12 ... 2014-11-21 19:54:16 :-)) 2014-11-21 19:54:21 musl and glibc both aim to conform to standards (c, posix, ieee math, etc.) 2014-11-21 19:54:25 bionic does not 2014-11-21 19:54:33 it's just random crap with the same names as the standard c functions 2014-11-21 19:55:02 katu: I don't like Google stuff. 2014-11-21 19:55:02 well, it works for its intended purpose, that is, completely custom runtime 2014-11-21 19:55:11 And I think they're worse than M$ stuff 2014-11-21 19:55:12 iirc google never claimed android is posix :) 2014-11-21 19:56:01 chatr: well .. NIH as any other megacorp. still, out of all NIH madness out there, google stuff tend to be of good quality. 2014-11-21 19:56:50 katu: But quality depends on what is your purpose 2014-11-21 19:57:13 If you wanna store every little thing on their servers 2014-11-21 19:57:17 that may be good 2014-11-21 19:57:30 well, in case of android 2014-11-21 19:57:37 they kinda bought it like this 2014-11-21 19:57:45 even the voice recognition is cloud based! 2014-11-21 19:57:49 That's insane 2014-11-21 19:57:51 :-( 2014-11-21 19:57:51 tried to correct their mistake with chromeos later, but it was too late 2014-11-21 19:58:00 android is THE mew PC 2014-11-21 19:58:06 But wait 2014-11-21 19:58:07 the unix which is good enough, despite there being plan 9 2014-11-21 19:58:09 etc 2014-11-21 19:58:28 It seems that we don't need the whole Linux apps compiled with Bionic 2014-11-21 19:58:36 to be able to run Android apps 2014-11-21 19:58:56 We should have some libraries 2014-11-21 19:59:00 including Bionic 2014-11-21 19:59:06 to be able to run Android apps 2014-11-21 19:59:10 That's what I think 2014-11-21 20:00:24 What do you think? 2014-11-21 20:06:46 hmm 2014-11-21 20:06:54 I'm reading this 2014-11-21 20:07:05 http://elinux.org/images/2/25/2013_elc_gentoo_bionic.pdf 2014-11-21 20:07:17 It says the main rationale for bionic 2014-11-21 20:07:24 was the licensing of apps 2014-11-21 21:11:07 Thank you guys 2014-11-21 21:11:13 bye 2014-11-21 23:02:07 Hi 2014-11-21 23:02:22 how can I disable framebuffer in Alpine Linux 2014-11-21 23:03:25 sometimes it causes problems 2014-11-21 23:04:50 I prefer 80x25 text mode 2014-11-21 23:08:08 any idea? 2014-11-21 23:08:12 I wanna set video=off 2014-11-21 23:08:15 or something like this 2014-11-21 23:08:31 in kernel parameters 2014-11-21 23:11:28 if you're using X you really should be using fb. modern drivers use kms (kernel modesetting) 2014-11-21 23:12:26 it doesn't come up 2014-11-21 23:12:30 on some systems 2014-11-21 23:12:47 So is there a way to change this? 2014-11-21 23:12:57 I used to set vga=xx 2014-11-21 23:13:00 or so 2014-11-21 23:13:11 in the boot time 2014-11-21 23:13:16 But in Alpine? 2014-11-21 23:14:20 I don't know much about the boot process. 2014-11-21 23:16:26 How can I change the resolution of the framebuffer 2014-11-21 23:16:29 Is it possible? 2014-11-21 23:17:06 the fbset utility probably can 2014-11-21 23:17:18 but nowadays isn't everyone using lcds? 2014-11-21 23:17:29 in which case there's only one correct resolution to run at -- the native one 2014-11-21 23:17:57 yeah 2014-11-21 23:18:07 but the problems is the tiny font 2014-11-21 23:19:49 :-p 2014-11-21 23:20:01 can fbcon use larger fonts than 8x16? 2014-11-21 23:21:32 maybe 2014-11-21 23:21:35 fbterm is great 2014-11-21 23:21:47 but it's not available in Alpine 2014-11-21 23:22:24 fbcon is also not available 2014-11-21 23:23:37 fbset says it can't find /etc/fb.modes 2014-11-21 23:28:03 I've downloaded fbset.deb from Ubuntu 2014-11-21 23:28:15 But I need 'ar' to extract it 2014-11-21 23:28:22 and I can't find it in Alpine 2014-11-21 23:28:45 any chance that you know which package contains it? 2014-11-21 23:28:57 or a command that searches to find out where it can be found 2014-11-21 23:38:27 I used some samples 2014-11-21 23:38:31 but they didn't work 2014-11-21 23:38:39 it seems that it is not easy to set fb parameters 2014-11-21 23:40:31 I'll be back 2014-11-21 23:44:14 vga=xx didn't have any impact on framebuffer resolution 2014-11-21 23:44:45 I should disable framebuffer 2014-11-21 23:58:10 damn 2014-11-21 23:58:17 https://bugzilla.redhat.com/show_bug.cgi?id=146003 2014-11-21 23:58:18 :-/ 2014-11-21 23:58:38 A kernel recompilation is required 2014-11-21 23:58:40 it seems 2014-11-22 00:20:55 At last I could be able to use 'fbterm 640x480-60' 2014-11-22 00:21:01 but the result is funny 2014-11-22 00:21:17 It uses a portion of screen 2014-11-22 00:21:23 to achieve 640x480! 2014-11-22 00:36:46 haha 2014-11-22 00:38:20 I'm compiling fbterm 2014-11-22 00:53:26 In fbterm 2014-11-22 00:53:40 some system calls from kernel is used 2014-11-22 00:53:45 what should I install 2014-11-22 00:53:50 to compile it correctly? 2014-11-22 01:20:08 done! 2014-11-22 01:20:56 :-) 2014-11-22 01:21:02 and now I can set font size 2014-11-22 01:21:06 whatever I want! 2014-11-22 01:21:43 Some minor changes were needed 2014-11-22 01:21:45 to compile this 2014-11-22 01:21:57 Adding some missed headers 2014-11-22 01:30:05 How can I contribute a package compiled? 2014-11-22 03:13:24 does alpine linux have gpg signed downloads available? 2014-11-22 03:14:07 guh, bad timing. going afk for an hour 2014-11-22 06:28:13 hm, how would i get tor working on alpine? i read somewhere that it was available in edge, but i mounted the edge iso and could not find the file 2014-11-22 07:27:18 ryonaloli, my guess is that it isn't on the edge iso 2014-11-22 07:27:26 ryonaloli, but should be in the edge repo 2014-11-22 07:28:45 ryonaloli, it is in edge/testing http://nl.alpinelinux.org/alpine/edge/testing/x86_64/ 2014-11-22 07:30:30 ryonaloli, it is also in http://nl.alpinelinux.org/alpine/latest-stable/testing/x86_64/ 2014-11-22 07:42:01 oh, thanks! 2014-11-22 07:58:54 ryonaloli, you should check out https://github.com/grugq/portal for reference of a good tor setup 2014-11-22 08:15:21 ryonaloli, do you have any particular goals with tor? 2014-11-22 08:16:18 yes, this'll be to ssh into a computer who's ip will change fairly often, and i'd like it to remain anonymous 2014-11-22 08:16:28 really all it'll be doing is mapping port 22 to localhost's 22 2014-11-22 08:16:38 for a tor hidden service, that is 2014-11-22 08:17:44 ryonaloli, I have never set that up before. Document on the wiki :P 2014-11-22 10:28:11 does alpine linux have a gpg signed hash? 2014-11-22 10:37:37 ryonaloli, I don't know for sure but I believe they are in /etc/apk/keys/ 2014-11-22 10:37:56 I don't know if there is a key on the website though 2014-11-22 10:38:15 *main website 2014-11-22 10:44:33 they're all RSA 1024 O_o 2014-11-22 10:44:33 i meant for the iso though 2014-11-22 10:52:27 not that I know of off the top of my head 2014-11-22 10:52:29 ncopa, were would one find keys for ISO releases and packages on the main website? 2014-11-22 10:52:29 ncopa, also is there any plans on moving to 2048 RSA keys or an Elliptic Curve algorithm keys? 2014-11-22 10:52:30 ryonaloli, nice for pointing that out I just assumed they wouldn't be 1024 keys 2014-11-22 10:52:32 i don't think there are any EC algorithms for any pgp applications. the closest i can think of is openbsd's signify program which uses ed25519 2014-11-22 10:52:33 ryonaloli, ughh seriously GPG 2014-11-22 10:52:34 it's the pgp standard's fault 2014-11-22 10:53:15 iirc all it supports are RSA, DSA, and ElGamal 2014-11-22 10:53:31 well I have issues with some of the choices of GPG but w/e 2014-11-22 10:53:53 like what? 2014-11-22 10:54:04 1. they will not allow copy and paste into GPG key for the graphical input agent or w/e 2014-11-22 10:55:24 2. claim there is no difference between 2048 and 4096 RSA 2014-11-22 10:55:46 so for 1 if you are using GPG for email such as I believe thunderbird and you have a pasword on your key they make the argument it is less secure to allow copy and paste 2014-11-22 10:56:17 which while potentially technically true the net security effect is worse 2014-11-22 10:56:36 point 1 i fully agree with. that's why i try not to use gpg-agent and just vanilla gpg. point 2 is kind of true. 2048 is good for a few more decades against the general field sieve algorithm, and both are useless against shor's quantum algorithm 2014-11-22 10:56:40 and generally if you can gain access to the copy/paste buffer in X you are generally hosed on maintaining security on that system 2014-11-22 10:56:46 but yeah the copy-paste thing for gpg-agent is pretty obnoxious 2014-11-22 10:58:26 for 2, I don't keep that up to date on crypto but all existing publicly accessible crypto algorithms aren't resistant to the quantum computing 2014-11-22 10:58:34 well, there is NTRU which has implementations, but it has a mild patent on it and it lacks forward secrecy, and there's also SIDH which is better in every way, but no implementations have been made yet 2014-11-22 10:59:04 SIDH is probably useless for gpg's purposes though, because it's based on diffie-hellman 2014-11-22 11:00:16 pgp 2.1 does ed25519 2014-11-22 11:00:21 YX9KwoR6rk2l: the program or standard? 2014-11-22 11:00:39 the standard is rfc4880 2014-11-22 11:00:57 ooh, sorry i meant gpg 2.1 2014-11-22 11:00:58 ooh 2014-11-22 11:00:58 oh that's awesome! 2014-11-22 11:01:13 also i have tor running without problems on my hosts 2014-11-22 11:13:53 ryonaloli, sent an email about it 2014-11-22 15:07:38 <^7heo> could someone put my hhpc port into testing? 2014-11-22 17:49:57 is there a man page for apk? I don't see an apk-doc package... 2014-11-22 17:50:55 also, alpine is awesome :) 2014-11-22 17:51:31 it feels like linux *should* feel - simple & lightweight 2014-11-22 17:51:43 ACTION glares at RHEL7 2014-11-22 18:13:33 <^7heo> RH is shit and has always been. 2014-11-22 18:13:36 <^7heo> rpm is shit. 2014-11-22 18:14:10 :) 2014-11-22 18:15:21 eh, I didn't mind it too much until v7 2014-11-22 18:16:38 using networkmanager (sorry, "NetworkManager") by default on server installs is beyond stupid 2014-11-22 18:17:43 haha yes 2014-11-22 18:18:46 on my laptop, sure. It usually handles wifi ok. But a server? wtf? 2014-11-22 19:41:36 hmm, apk info -W doesn't seem to handle symlinks..? 2014-11-22 19:41:45 busybox, in this case 2014-11-22 19:43:03 I guess because the symlinks aren't "in" the package? 2014-11-22 20:08:32 hm, why would procps's top not show cpu usage but busybox's top does? 2014-11-22 20:16:16 lol, I just installed Plone (a "huge" CMS) in alpine 2014-11-22 20:16:30 glad it didn't barf on musl 2014-11-22 23:11:09 i'm unable to boot alpine on my computer because of https://bugs.alpinelinux.org/issues/2616 2014-11-22 23:12:18 it just freezes at "Loading Hardware Drivers" and does not pass. i've tried both the x86 and x86_64 versions, tried booting with nomodeset and pax_nouderef, with no luck 2014-11-22 23:13:29 i do not have a serial cable, but if i boot with console=ttyS0,9600, wait a minute or two, then blindly type "root", , "reboot", , it does power off, so the problem is related to the display 2014-11-22 23:23:21 the bug report was made 10 months ago and still has the "new" status and no replies ;_; 2014-11-22 23:35:29 <^7heo> I never understood the point of networkmanager even for WiFi 2014-11-22 23:35:45 <^7heo> I mean, unless you really NEED the priority on WiFi 2014-11-22 23:36:10 <^7heo> (i.e. selection of one network among multiple networks available SIMULTANEOUSLY) 2014-11-22 23:36:24 <^7heo> you can be perfectly fine with just wpa_supplicant... 2014-11-22 23:36:27 burning vanilla and about to test it out. if i don't come back after quitting i probably gave up and used a debian live cd 2014-11-22 23:36:35 (short on time) 2014-11-23 01:14:15 <^7heo> Okay so it seems that vanilla isn't the solution to that problem. 2014-11-23 01:14:24 <^7heo> To be perfectly honest, I expected so. 2014-11-23 01:15:20 <^7heo> However 2014-11-23 01:15:39 <^7heo> i do not have a serial cable, but if i boot with console=ttyS0,9600, wait a minute or two, then blindly type "root", , "reboot", , it does power off, so the problem is related to the display 2014-11-23 01:15:56 <^7heo> That is of upmost interest. 2014-11-23 01:17:01 <^7heo> And the fact that a bug report for this has been made 10 months ago, and still has no reply and no activity shows one thing: it's not happening often. 2014-11-23 01:17:13 <^7heo> However, it did happen to me. 2014-11-23 01:17:22 <^7heo> With the laptop of my flatmate. 2014-11-23 01:17:37 <^7heo> So I'll try to reproduce it and search what I acn find. 2014-11-23 01:17:44 <^7heo> can* 2014-11-23 08:45:04 Does someone here use st? 2014-11-23 08:58:00 héhé! today should the D day jor me ^^ normally my hands-and-eyes is going to my Data-center to plug 6 alpine USB key is rack of 10 machines ;) I will then proceed to the update and install of our ststem on top of it. All without interrupting the usage (but for the time of the reboot on the usb KEY end a couple of minutes to restart services in KVMs...) 2014-11-23 08:58:20 the switch is from debian to alpine 2014-11-23 08:58:46 my first alpine in production ;) 2014-11-23 08:59:29 \o/ 2014-11-23 20:53:51 i am running alpine-xen-3.0.6-x86_64 as the host os and have some quistions about using dom0 or a domu 2014-11-23 20:54:07 this host os is running of a usb stick 2014-11-23 20:54:42 i want to install a dhcpd and caching name server of some sorts for use by my centos domu's 2014-11-23 20:55:48 my question should i install a dhcpd for use with my physical lan and DOMu's in dom0, or create another DOMu and install the dhcpd in there? 2014-11-23 20:56:16 also, i want to install a http caching server to cache packages downloaded by yum and also web packages 2014-11-23 20:56:33 again, should i install this caching server in a dom0 or in a domu ? 2014-11-23 20:56:45 with regards to performance 2014-11-23 20:58:02 my domu's runs of a LV which is made up by hard drives 2014-11-23 20:58:23 so domu io will be faster because hard drives are faster than USB devices 2014-11-23 21:07:53 alpine-linux-use: i have a separate vm to provide services like that 2014-11-23 21:08:29 it's a bit cleaner than putting the services in dom0, just need to make sure that one starts a bit earlier than the others 2014-11-23 21:08:48 i see thanks. 2014-11-23 21:10:12 I will then want to install the smallest possible OS for a new DOMU for a caching server and also a dhcpd server that will be used by my physical LAN and other DOMU's 2014-11-23 21:10:31 iv'e been enjoying Centos 2014-11-23 21:11:13 But am looking to rather install Alpine as this DOMU's OS 2014-11-23 21:12:00 Do you run your VM's off LV's or img files? 2014-11-23 21:14:50 lvm 2014-11-23 21:15:04 lv's, sorry :) 2014-11-23 21:15:17 i understand 2014-11-23 21:16:59 but how about installing the dhcpd in DOMO but have it run completely in ram like /dev/shm somewhere, in stead of it's startup scripts and config scripts which will stay inside /etc somewhere 2014-11-23 21:17:55 but how about installing the dhcpd in DOMO but have it run completely in ram like /dev/shm somewhere in stead, while keeping it's startup scripts and config files in /etc somwhere 2014-11-23 21:18:05 you know that you can run alpine in ram completely? 2014-11-23 21:18:13 no need to do it for a single daemon 2014-11-23 21:18:30 i don't have such a setup here, but it's possible and builtin on alpine 2014-11-23 21:18:58 i see. 2014-11-23 21:19:49 so Alpine-Linux (being the hosting OS and DOM0) can run completely in RAM?> 2014-11-23 21:19:54 ? * 2014-11-23 21:21:04 and then my LV's runs off LV's 2014-11-23 21:21:22 then my actual question is this 2014-11-23 21:21:39 now the USB disk crashed 2014-11-23 21:21:53 so then I can make a new bootable alpine linux hosting OS with DOM0 2014-11-23 21:22:24 on a new USB disk 2014-11-23 21:22:35 plug that in 2014-11-23 21:23:00 but how difficult is it to tell it to pick up by still existing LVG so it can access all the VM's on it 2014-11-23 21:27:42 i meant is it easy to configure the new USB disk running the new Alpine Linux DOM0 to pick up and use the LVG that was created with the previous USB disk that was running the previous Alpine version and DOM0 but have now crashed for example 2014-11-23 21:28:28 so i can access my DOMU's again which lived in LV's inside the LVG 2014-11-23 21:45:48 alpine-linux-use: at worst it'd be pvscan && vgscan && vgchange -ay i think 2014-11-23 21:46:03 try it before you try to build workaroudns 2014-11-23 21:46:08 or design 2014-11-23 21:47:08 yes thanks, looks simple enough 2014-11-23 21:48:14 il try those again, just copying /etc/xen off :) 2014-11-23 22:12:28 ncopa, someone should ping me when @alpinelinux has news to tweet and i'll RT. somehow i missed a few. 2014-11-23 22:48:20 darkfader: it game me memory errors, but worked, i can console 'log' in to the domu 2014-11-23 22:48:54 libxl: error: libxl_device.c:274:libxl__device_disk_set_backend: Disk vdev=sda failed to stat: /dev/vg0/ns1 localhost:/etc/xen# pvscan && vgscan && vgchange -ay PV /dev/sda VG vg0 lvm2 [232.88 GiB / 166.70 GiB free] PV /dev/sdb VG vg0 lvm2 [279.46 GiB / 279.46 GiB free] Total: 2 [512.34 GiB] / in use: 2 [512.34 GiB] / in no VG: 0 [0 ] Reading all physical volumes. This may take a while... Found vo 2014-11-23 22:49:10 :( 2014-11-24 08:03:28 <^7heo> http://dilbert.com/dyn/str_strip/000000000/00000000/0000000/100000/60000/0000/400/160498/160498.strip.gif < How I feel work is. 2014-11-24 08:27:23 morning 2014-11-24 08:28:07 <^7heo> moin 2014-11-24 08:35:04 <^7heo> ncopa: do you think you could push hhpc in the repos? 2014-11-24 09:04:24 dalias: ok, I'll ping you. I'm just not that much in to twitter 2014-11-24 09:04:42 i suppose I need to figure out what/when to tweet 2014-11-24 09:05:17 ^7heo: the funny thing about dilbert is that most tech ppl reads it as a satire 2014-11-24 09:05:36 while managers reads it as "oh this was actually a great idea" 2014-11-24 09:05:51 <^7heo> ncopa: was it ever meant as something else than a satire? 2014-11-24 09:05:57 no 2014-11-24 09:06:11 <^7heo> you had me doubting at some point. 2014-11-24 09:06:12 the funny thing is that managers often does not read it as satire :) 2014-11-24 09:06:14 <^7heo> yeah 2014-11-24 09:06:22 <^7heo> that's just added fun. 2014-11-24 09:06:26 that makes dilbert double fun :) 2014-11-24 09:06:50 <^7heo> I am happy to see that we agree on that :P 2014-11-24 09:07:27 <^7heo> ncopa: do you think you could please put hhpc in testing? :) 2014-11-24 09:17:53 ^7heo: didnt i already do that? 2014-11-24 09:18:20 hmm? apparently not? 2014-11-24 09:19:31 http://git.alpinelinux.org/cgit/aports/commit/?id=c1abc6379217f1543652e2b4651eb61e88801246 2014-11-24 09:20:46 <^7heo> ncopa: thanks ;) 2014-11-24 09:44:34 ncopa: what's your advice for upgrading 2.7 to 3.0? I'm going to try again 2014-11-24 09:44:45 apk update && apk upgrade --no-selfupgrade ? 2014-11-24 09:45:45 and then apk upgrade --update-cache --available ? 2014-11-24 09:47:35 i'd say start with apk add busybox-static, before you chagne any repos 2014-11-24 09:47:50 :) 2014-11-24 09:48:02 http://wiki.alpinelinux.org/wiki/Upgrading_Alpine#Upgrading_Alpine_v2.x_to_v3.x 2014-11-24 09:49:43 ah, thanks for the link 2014-11-24 10:19:44 so 2014-11-24 10:19:46 apk add busybox-static apk-tools-static 2014-11-24 10:20:00 apk.static update && apk.static upgrade --no-self-upgrade --available 2014-11-24 10:20:02 worked 2014-11-24 10:20:12 what next? 2014-11-24 10:21:13 apk del apk-tools-static busybox-static && apk upgrade --available ? 2014-11-24 10:44:18 if you got any errors, then you can do apk fix 2014-11-24 10:44:21 if no errors 2014-11-24 10:44:25 then you are done 2014-11-24 10:44:41 you can keep the -static packages for future emergencies if you want 2014-11-24 10:44:55 otherwise, apk del apk-tools-static busybox-static 2014-11-24 10:50:38 i can apk del them, but they will still be in /etc/apk/cache 2014-11-24 10:50:55 so that's good 2014-11-24 10:54:45 ncopa: how can i update the packages on a USB stick in /media/usb/apks/x86_64/ instead of in apkvol.tar.gz ? 2014-11-24 14:07:19 nice, http://www.extremetech.com/extreme/192929-255tbps-worlds-fastest-network-could-carry-all-the-internet-traffic-single-fiber 2014-11-24 14:08:28 I wonder what is the upper limit, when they say. "but we will eventually run up against the laws of physics" 2014-11-24 15:08:25 <^7heo> ncopa: is it just me, or lvm2-doc is empty? 2014-11-24 15:08:35 <^7heo> $ apk info -L lvm2-doc 2014-11-24 15:08:35 <^7heo> lvm2-doc-2.02.112-r0 contains: 2014-11-24 15:08:37 <^7heo> $ 2014-11-24 15:08:55 ^7heo, you need package installed for that to work 2014-11-24 15:09:25 <^7heo> $ apk version lvm2-doc 2014-11-24 15:09:25 <^7heo> Installed: Available: 2014-11-24 15:09:25 <^7heo> lvm2-doc-2.02.112-r0 = 2.02.112-r0 2014-11-24 15:09:26 <^7heo> $ 2014-11-24 15:09:37 <^7heo> better? 2014-11-24 15:34:20 i have lvm2-doc-2.02.106-r0 and it's fine. are you in edge? 2014-11-24 15:34:46 checking edge 2014-11-24 15:35:37 main is 2.02.111 2014-11-24 15:36:27 alpine-dev-edge:~$ sudo apk add lvm2-doc 2014-11-24 15:36:27 (1/1) Installing lvm2-doc (2.02.112-r0) 2014-11-24 15:36:27 ERROR: lvm2-doc-2.02.112-r0: No error information 2014-11-24 15:42:00 <^7heo> ScrumpyJack: I'm on edge yes. 2014-11-24 15:42:51 <^7heo> btw, stable has the version you have 2014-11-24 15:43:19 where are the APKBUILDs for edge? 2014-11-24 15:46:49 <^7heo> in the git, I guess? 2014-11-24 15:46:53 <^7heo> not sure about that one. 2014-11-24 15:47:16 <^7heo> okay the lvm2-doc package is fine in stable. 2014-11-24 15:48:03 apk info -L lvm2-doc 2014-11-24 15:48:25 lvm2-doc-2.02.112-r1 contains: 2014-11-24 15:48:35 $ apk info -L lvm2-doc | sprunge 2014-11-24 15:48:35 http://sprunge.us/hKKK 2014-11-24 15:48:52 i am hacking tmux support for xendomains 2014-11-24 15:48:55 i think i got it working 2014-11-24 15:48:56 <^7heo> that's r1 2014-11-24 15:48:59 <^7heo> I got r0 2014-11-24 15:49:04 apk upgrade? 2014-11-24 15:49:06 <^7heo> ok 2014-11-24 15:49:08 <^7heo> I was going to ask. 2014-11-24 15:49:13 apk info -L lvm2-doc 2014-11-24 15:49:18 show contents of package 2014-11-24 15:49:50 <^7heo> I know too :) 2014-11-24 15:50:01 <^7heo> see 1h ago ;) 2014-11-24 15:50:17 ERROR: lvm2-doc-2.02.112-r0: No error information 2014-11-24 15:50:32 looks like index is out of date 2014-11-24 15:50:41 apk update 2014-11-24 15:50:49 <^7heo> ncopa: I confirm, fixed. 2014-11-24 15:59:40 ncopa: i have the main and testing git repros. where are the edge repros? 2014-11-24 16:02:38 edge/main and edge/testing 2014-11-24 16:02:43 if you have those then you have the edge repos 2014-11-24 16:02:45 gotta run 2014-11-24 16:02:46 see u 2014-11-24 16:02:47 and what the hell do you mean by tmux support for xen domains? do you mean open a new tmux window that connects to directly to a domain console by just supplying the domain ID? that sounds very very cool. 2014-11-24 16:03:09 yup 2014-11-24 16:03:27 xendomains initd script can start domUs in tmux windows 2014-11-24 16:03:39 so you can reach the xen domains console with tmux attach 2014-11-24 16:03:43 gotta run again 2014-11-24 16:03:44 see u 2014-11-24 16:04:27 speak soon - you have my attention with tmux+xen ;) 2014-11-24 16:18:22 ncopa, *nod*, i just think it would be useful for promoting alpine. 2014-11-24 16:18:50 ncopa, i have 110 followers on twitter and steadily increasing, and most are interested in this kind of thing 2014-11-24 16:23:09 dalias, is this musl related, http://bugs.alpinelinux.org/issues/3282 ? 2014-11-24 16:23:49 if yes, any possibility getting fixed before AL v3.1 release :-) 2014-11-24 16:25:59 cannot of get lots of coreutils command coz of it 2014-11-24 16:26:05 cannot get* 2014-11-24 16:36:57 vkrishn, utmp stubbing is intentional, but that can be discussed if there's really a reason you need it 2014-11-24 16:37:14 the classic utmp mechanism of direct file access with suid/sgid is idiotic and insecure tho 2014-11-24 16:37:41 so how to get command like `who` etc.. get working ? 2014-11-24 16:38:05 i wonder if there's a library replacement for the utmp interfaces to go thru some more secure mechanism 2014-11-24 16:38:07 any temprorary patch, which may not in musl main 2014-11-24 16:38:31 may not be included in musl, for now 2014-11-24 16:38:44 my view was always that they leak private information to all users (even unprivileged ones like a compromised nobody) and don't have significant positive use 2014-11-24 16:39:01 for auditing of logins you can check the syslog for sshd or similar 2014-11-24 16:39:09 rather than publishing that data to world 2014-11-24 16:39:28 so my view was always "it's better if they don't work" 2014-11-24 16:39:48 but if there's significant objection to this we should look for a solution of some sort that's satisfactory for ppl who want it 2014-11-24 16:39:52 either in musl or outside musl 2014-11-24 16:41:14 but is musl suppose to me security oriented or something that PAX/SSP should take care off ? 2014-11-24 16:41:39 or is it not POSIX / 2014-11-24 16:41:41 ? 2014-11-24 17:52:13 till now found 'users, who, last' command won't work ;) 2014-11-24 17:59:40 gtg, but would be checking irc logs 2014-11-24 19:28:09 Yay alpine docker containers 2014-11-24 22:07:36 Is it normal for OpenRC to "hang" for about 10-20 seconds, with the cursor just blinking, before it starts to load stuff? 2014-11-24 22:08:47 vm? real machine? 2014-11-24 22:09:22 (i havent seen such thing in qemu fwiw) 2014-11-24 22:09:38 real machine 2014-11-24 22:10:09 I have never used openrc before so I have no comparison. 2014-11-24 22:20:20 maybe dmesg or /var/log/messages can tell you what's going on 2014-11-24 22:20:54 iio7: i have that on my vm hosts, but not always and no idea what it's caused by 2014-11-24 22:21:22 i can tell on my oldest (233mhz) laptop it took too long to wait through :) 2014-11-24 22:22:56 Hmm, that's strange, I have never seen this behavior before except on Debian because there was a problem with udev, but that was sysinit and not openrc. 2014-11-24 22:23:38 openrc in alpine barfed on me for couple of things 2014-11-24 22:24:06 unexpected things in fstab, hostname not in /etc/hosts and broken resolver and some other incidents i think 2014-11-24 22:24:24 you can turn on openrc logging in /etc/rc.conf 2014-11-24 22:24:25 its fairly spartan to tell whats wrong :( 2014-11-24 22:24:31 ah 2014-11-24 22:24:46 but that probably wont help if the hang happens before things get mounted 2014-11-24 22:25:05 well, yeah 2014-11-24 22:25:19 but might help if looking from host at lxc container i suppose 2014-11-24 22:27:27 Will it create its own log file or will it log to messages? 2014-11-24 22:29:47 Okay, logging activated, reboot, it hangs.. lets see if something turns up 2014-11-24 22:30:34 I actually suspect it is hard drive related somehow. 2014-11-24 22:30:50 then you may want to turn on kernel debug things 2014-11-24 22:31:39 The rc.log doesn't reveal anything useful. 2014-11-24 22:33:23 I am seeing this in dmesg, http://pastebin.com/u6qTzhKz but not sure it is directly related 2014-11-24 23:38:53 BitL0G1c, hello 2014-11-25 01:21:49 would it be possible for alpine's firefox package to fix the intentional omission of h264 video support? 2014-11-25 03:49:05 dalias, or at least make it stupid simple to add it in 2014-11-25 03:49:45 iirc they just intentionally disable the ability to use the codec even though it's there, for stupid political reasons 2014-11-25 03:49:59 so i would think fixing this is just a one-line patch or something 2014-11-25 03:50:16 well I mean like linux mint's one click 2014-11-25 03:50:19 media pack install 2014-11-25 03:50:31 that sorta side steps that issue 2014-11-25 03:50:43 since it is technically not shipped with and user requests 2014-11-25 03:50:59 i guess for alpine one meta package or something 2014-11-25 03:51:12 dalias, oh btw do you have TPE enabled with grsec 2014-11-25 03:51:28 I am running into errors running Firefox and Midori 2014-11-25 03:51:31 they crash on me 2014-11-25 03:51:40 i don't use the grsec kernel at all 2014-11-25 03:51:42 it breaks too much 2014-11-25 03:52:03 cripples /proc to be useless; top can't even show how much memory my programs are using :( 2014-11-25 03:52:12 sorta the point 2014-11-25 03:52:29 you would use top from the grsec admin 2014-11-25 03:52:36 which you could enable or disable I believe 2014-11-25 03:52:42 but I don't know enough to give a good answer 2014-11-25 03:52:45 just a guess 2014-11-25 03:53:16 no, the files with the info are completely missing from /proc 2014-11-25 03:53:24 i don't think using grsec admin would help 2014-11-25 03:54:00 well it might be changed with sysctl 2014-11-25 03:54:06 i would need to go and look 2014-11-25 03:58:52 dalias, understandble that it breaks a lot 2014-11-25 03:59:11 just wish there was a stronger effort of compliance with grsec/pax and SELinux 2014-11-25 03:59:33 it would move Linux in a better direction for security 2014-11-25 08:16:18 <^7heo> moin 2014-11-25 09:13:14 morning 2014-11-25 11:33:18 hi 2014-11-25 11:33:44 When following http://wiki.alpinelinux.org/wiki/Qemu to install a brand new KVM based on alpine... 2014-11-25 11:33:53 Launching the KVM for the first time from the freshly downloaded ISO starts well but then proceed to switch to "1024 x 768 Graphic mode" which I cannot follow anymore in my terminal window :( 2014-11-25 11:34:16 My command to launch the KVM is 2014-11-25 11:34:26 qemu-system-x86_64 -enable-kvm -smp 1 -m 256 -cdrom alpine-mini-3.0.6-x86_64.iso -hda kvm-alpine.qcow2 -boot d -net nic -net user -nographic -localtime -display curses 2014-11-25 11:35:48 I had a graphical desktop in my lab but I have no graphic on my prod machines and they are 1000km away :( 2014-11-25 11:37:42 Jean-Scotch: tell qemu to use vnc and attach to it from a graphical desktop 2014-11-25 11:38:53 VNC accross the internet from 1000km away? I do not really like it as a production access path for massive deployments :/ 2014-11-25 11:39:37 this is the first one but it's planned to have a lot of those. And they should be automated soon 2014-11-25 11:53:43 "1024 x 768 Graphic mode"? 2014-11-25 11:54:05 yes :( that what's displayed in the middle of my term 2014-11-25 11:54:48 it's a gnome term on awesome desktop connected via ssh to a screen in CLI on the remote server 2014-11-25 11:55:54 the remote server is the host running the KVMs? what linux? 2014-11-25 11:57:00 Jean-Scotch: I think you should be able to use the virtio-serial console, but I have no idea how to enable that 2014-11-25 11:57:33 the host is a debian wheezy 2014-11-25 11:57:44 Jean-Scotch: or the emulated Qemu serial port 2014-11-25 11:57:46 with recent qemu and openvswitch 2014-11-25 11:58:20 it's the last one standing all other machines in the rack are being switched to alpine ;) 2014-11-25 11:58:38 sorry, can't really help you with KVM :( 2014-11-25 11:58:42 but need some spare services to run during migration... 2014-11-25 12:01:24 I want to start several of those kvm-alpine to hosts an openvpn (14 actives vpns) and a nginx to start with 2014-11-25 12:02:58 Jean-Scotch: I would set one using vnc, then make it boot to serial and cp/dd the disk image in order to replicate it 2014-11-25 12:06:39 i have a bunch on alpine, ubuntu, debian and sles hvm and pv xen domUs running on alpine-xen (alpine dom0) . works like a charm 2014-11-25 12:06:52 perhaps move from KVM to xen? :) 2014-11-25 12:20:06 Jean-Scotch: I use qemu for testing kernels with -nographics 2014-11-25 12:20:22 the problem is kms 2014-11-25 12:20:55 kms? 2014-11-25 12:21:09 kernel mode setting 2014-11-25 12:21:16 ok 2014-11-25 12:21:32 let me see my setup... 2014-11-25 12:21:38 do you see the boot: prompt? 2014-11-25 12:22:01 the graphical switch is after seeing the start of the alpine boot sequence 2014-11-25 12:23:08 if you see the "boot:" then i htink you should be able to type in nomodeset or similar 2014-11-25 12:23:14 grsec nomodeset 2014-11-25 12:24:01 this is what i use on remote ssh: qemu-system-x86_64 -smp 4 --enable-kvm -m 2048 -curses ... 2014-11-25 12:24:17 no, i see nothing. juste the report of the end of the alpine boot sequence 2014-11-25 12:24:54 for the start it's the same command I use 2014-11-25 12:25:24 qemu-system-x86_64 -serial stdio -smp 4 --enable-kvm -m 2048 -curses -cdrom alpine-edge.iso -boot d 2014-11-25 12:25:31 I see the boot prompt. sorry 2014-11-25 12:25:41 i dont think the -serial i use is related 2014-11-25 12:26:06 which iso are you booting? 2014-11-25 12:26:34 mini 2014-11-25 12:26:36 sorry 2014-11-25 12:26:37 ok 2014-11-25 12:26:38 the lastest mini 2014-11-25 12:27:11 I'll try here 2014-11-25 12:28:26 I can type commands at boot but nomodeset is not a valid one 2014-11-25 12:29:10 when you see the boot prompt, quicly press so it stop booting 2014-11-25 12:29:17 then you type: grsec nomodeset 2014-11-25 12:29:29 Welcome to Alpine Linux 3.0 2014-11-25 12:29:29 Kernel 3.14.22-1-grsec on an x86_64 (/dev/tty1) 2014-11-25 12:29:58 oh i know what happens 2014-11-25 12:30:00 yep! thanks. 2014-11-25 12:30:09 it loads the kernel module later during boot 2014-11-25 12:30:18 but later on will it need that to start in the background? 2014-11-25 12:30:42 i think the only qemu manager we have i libvirt 2014-11-25 12:30:47 wich is kinda heavy 2014-11-25 12:30:54 and you'll have to manually edit xml 2014-11-25 12:30:57 I prefer to do it manually 2014-11-25 12:31:05 with my own scripts 2014-11-25 12:31:15 then, after install to qcow disk 2014-11-25 12:31:19 before reboot 2014-11-25 12:32:01 mount the disk and edit etc/update-extlinux.conf 2014-11-25 12:32:09 you can specify extra boot options there 2014-11-25 12:32:38 default_kernel_opts="quiet nomodeset" 2014-11-25 12:33:09 would probably be nice to have an option for setup-disk that can set that 2014-11-25 12:33:43 or do it automatically when qemu 2014-11-25 12:33:54 but i also have graphics qemu 2014-11-25 12:38:12 ok. I'm at the stage of editing etc/update-extlinux.conf 2014-11-25 12:38:42 i think you might need edit boot/extlinux.conf too first boot 2014-11-25 12:39:24 and after boot: apk fix syslinux 2014-11-25 12:39:31 which should trigger the update-extlinux 2014-11-25 12:40:00 alternatively, after editing update-extlinux: apk fix --root /mounted-root syslinux 2014-11-25 12:40:01 what to change in boot/extlinux.conf ? 2014-11-25 12:40:30 add nomodeset to APPEND 2014-11-25 12:40:34 after quiet 2014-11-25 12:41:00 there is no APPEND in boot/extlinux.conf... 2014-11-25 12:41:15 LABEL grsec 2014-11-25 12:41:15 MENU DEFAULT 2014-11-25 12:41:15 MENU LABEL Linux grsec 2014-11-25 12:41:15 LINUX vmlinuz-grsec 2014-11-25 12:41:15 INITRD initramfs-grsec 2014-11-25 12:41:16 APPEND root=UUID=b5bf0e87-e936-40d0-9ba8-b7fbb8c63245 modules=sd-mod,usb-storage,ext3,ext4 quiet 2014-11-25 12:41:18 is what i have 2014-11-25 12:41:43 hm 2014-11-25 12:43:25 maybe we should disable modeset by default when using QEMU disk 2014-11-25 12:43:39 its kinda inconvenient to disable it install time 2014-11-25 12:51:26 i suppose the simplest is to at first reboot, press again at boot, then manually add 'nomodeset' kernel parameter 2014-11-25 12:51:45 and after bootup, edit /etc/update-extlinux.conf and apk fix syslinux 2014-11-25 12:55:03 at first boot from qcow disk, press at boot menu and append 'nomodeset': http://imgur.com/JtRmYnb 2014-11-25 12:56:11 after bootup, edit /etc/update-extlinux.conf 2014-11-25 12:56:14 default_kernel_opts="quiet nomodeset" 2014-11-25 12:56:22 apk fix syslinux 2014-11-25 12:56:28 will try now but was delayed by a phone call 2014-11-25 12:56:29 to regerneate the /boot/extlinux.conf 2014-11-25 12:56:34 and then reboot 2014-11-25 12:56:44 ..just to verify that it worked 2014-11-25 12:56:52 worked here 2014-11-25 12:56:59 inconvenient though... 2014-11-25 12:57:03 my current extlinux.conf file is # Generated by update-extlinux 6.03_pre19-r0 2014-11-25 12:57:04 DEFAULT menu.c32 2014-11-25 12:57:04 PROMPT 0 2014-11-25 12:57:04 MENU TITLE Alpine/Linux Boot Menu 2014-11-25 12:57:04 MENU HIDDEN 2014-11-25 12:57:04 MENU AUTOBOOT Alpine will be booted automatically in # seconds. 2014-11-25 12:57:04 TIMEOUT 30 2014-11-25 12:57:05 MENU SEPARATOR 2014-11-25 12:57:40 hm? 2014-11-25 12:58:27 Jean-Scotch: is it a sysinstall? 2014-11-25 12:58:36 maybe a first reboot first... 2014-11-25 12:58:41 or / is tmpfs? 2014-11-25 12:59:01 it's the standard install (first pass) from iso to qcow2 2014-11-25 12:59:13 just run setup-alpine 2014-11-25 12:59:19 nothing else yet 2014-11-25 12:59:27 not even rebooted 2014-11-25 12:59:40 and on the disk question, 'sys' 'data' or 'tmpfs' what did you answer? 2014-11-25 12:59:51 sys 2014-11-25 12:59:53 to sda 2014-11-25 12:59:59 ok 2014-11-25 13:00:24 then you should be able to poweroff and at first boot press again 2014-11-25 13:00:31 need to reboot to start from sda now 2014-11-25 13:00:37 yes 2014-11-25 13:00:47 dot his at the first boot: http://imgur.com/JtRmYnb 2014-11-25 13:00:56 do this* 2014-11-25 13:13:40 I start again from scratch to be able to properly reproduce 2014-11-25 13:18:27 speaking of qemu, ncopa, http://sprunge.us/FOOg 2014-11-25 13:18:36 are all the deps really needed? 2014-11-25 13:18:45 yes. the defaut boot menu for a fresh install in KVM qcow2 is empty! 2014-11-25 13:19:59 Jean-Scotch: thats weird 2014-11-25 13:20:13 ScrumpyJack: we have support for gtk, which pulls in those deps 2014-11-25 13:34:17 first lesson: do not use a to small qcow file! use 512M or more... 2014-11-25 13:43:17 ncopa: with your modifs and a little bit space more for the qcow, it seems to be working... 2014-11-25 14:20:52 why does qemu need gtk? 2014-11-25 14:39:47 ScrumpyJack: it has a gtk interface 2014-11-25 14:39:59 imho, it would been nice if it was implemented as a plugin 2014-11-25 15:04:15 is the startup script for openvpn designed to start only one instance of openvpn "openvpn.conf" or is it possible without hacking the startup scripts to launch several instances? 2014-11-25 15:04:49 just adding *.conf files doesn't work 2014-11-25 15:05:15 Jean-Scotch: iirc, it supports both 2014-11-25 15:05:34 i think you can: ln -s openvpn /etc/init.d/openvpn.myinstance 2014-11-25 15:05:51 and config it in /etc/conf.d/openvpn.myinstance 2014-11-25 15:06:16 iirc, it will look for /etc/openvpn/myinstance.conf 2014-11-25 15:08:57 I was hoping to just throw defenition files in some etc subdir and restart openvpn :( not to edit some registry key 2014-11-25 15:09:38 I will change the starup script... 2014-11-25 15:13:43 ncopa: what gtk interface? 2014-11-25 15:14:25 ScrumpyJack: this: https://www.youtube.com/watch?v=myyEss6ErG8 2014-11-25 15:14:46 perhaps it's to create an X window? 2014-11-25 15:15:54 ah i see 2014-11-25 15:17:15 (no flash here, but i get the title) 2014-11-25 15:17:21 is it this? http://gtk-qemu.sourceforge.net/ 2014-11-25 15:18:37 no 2014-11-25 15:18:57 you can run qemu in a gtk window 2014-11-25 15:19:02 >>>> Since this is a pretty visible change for a lot of people, I thought I'd 2014-11-25 15:19:05 >>>> send a top level note. The GTK UI is now committed and is the default 2014-11-25 15:19:08 >>>> UI provided it's available. 2014-11-25 15:19:12 :( 2014-11-25 15:19:46 surely it's optional 2014-11-25 15:19:52 yes, compile time 2014-11-25 15:20:10 i suppose we could compile one binary qemu and another qemu-gtk 2014-11-25 15:20:21 :) yay! so we can move it into another package? 2014-11-25 15:20:25 but i dont like that 2014-11-25 15:20:35 i'd prefer to have it as a plugin 2014-11-25 15:21:29 but it's not in the standard default base qemu-i386 in alpine-xen. if you want another emulation, you need so much bloat :( 2014-11-25 15:21:31 i do agree it would be nice to not include it in default qemu pkg 2014-11-25 15:21:57 qemu in xen was not compiled with gtk 2014-11-25 15:21:57 :) 2014-11-25 15:23:55 so you can imagine how annoying it is when all you want is qemu-system-something in a dom0, you end up with an extra 227Mb of useless packages 2014-11-25 15:25:13 shall i add something on b.a.o? 2014-11-25 15:58:27 sure 2014-11-25 15:58:30 so we dont forget 2014-11-25 15:58:57 maybe also ask qemu devs if its possible to implement it as plugin 2014-11-25 15:59:03 gtk support as plugin 2014-11-25 15:59:15 hi 2014-11-25 16:00:17 hi and bye for today :) 2014-11-25 16:02:07 <^7heo> ncopa: o/ 2014-11-25 16:07:27 Is there a way to handle APKBUILD with git version without having to use a static source url ? 2014-11-25 16:08:50 The example on the wiki suggest to make an archive and then sending it to some url (or maybe to the startdir) 2014-11-25 16:09:47 I like the way makepkg permits to simply use a git url (sorry if the comparison annoy anyone) 2014-11-25 16:09:54 Is it possible ? 2014-11-25 16:11:23 quinq: http://wiki.alpinelinux.org/wiki/APKBUILD_examples:Git_checkout 2014-11-25 16:11:49 Yes, that what I was talking about 2014-11-25 16:11:58 oh ,sorry 2014-11-25 16:11:59 *that's 2014-11-25 16:12:04 No problem :) 2014-11-25 16:19:43 I have looking APKBUILDs that use git as source and all of them make an archive, so I *think* that it is not possible :( 2014-11-25 16:20:07 Well I also just looked ^^ 2014-11-25 16:20:29 It seems that the snapsho() function has been moved to abuild 2014-11-25 16:20:40 So there's only to define urls now 2014-11-25 16:21:23 I manage to fetch a snapshot to startdir (with disturl="$startdir") 2014-11-25 16:21:54 great 2014-11-25 16:22:08 But it can't find the file for checksum with source="$startdir/$pkgname-$pkgver.tar.gz" 2014-11-25 16:22:28 (filename is correct, but it seems that it doesn't take the $startdir in account) 2014-11-25 16:24:07 '$startdir' is where APKBUILD resides, is not? 2014-11-25 16:25:57 It is 2014-11-25 16:26:06 (and the snapshot) 2014-11-25 16:26:24 But removing $startdir frome source resolved the problem 2014-11-25 16:26:56 ah, cool 2014-11-25 16:27:19 It automatically links src/$source to $source 2014-11-25 16:27:39 and then calls sha512sum on src/$source 2014-11-25 16:27:46 ok, thanks for the help :) 2014-11-25 16:29:15 you helped yourself more than me ;) 2014-11-25 16:42:29 Well now it won't build because it can't find source in /var/cache/distfiles 2014-11-25 16:43:04 I should have a look at abuild sources 2014-11-25 16:49:39 There are some issues 2014-11-25 16:49:52 I would like to be able to fille the gap... http://wiki.alpinelinux.org/wiki/Setting_up_a_OpenVPN_server#More_than_one_server_or_client 2014-11-25 16:50:26 snapshot() creates an archive in src by default, then makeshum() erases that file with a symlink pointing to $startdir 2014-11-25 17:09:25 Maybe they should use SRCDEST 2014-11-25 17:09:57 Because then in default_unpack, it arbitrarely sets the source to $SRCDEST/$source 2014-11-25 17:59:16 i have a question about setting up bridges in 3.0.6 2014-11-25 17:59:47 i do 2014-11-25 17:59:58 service networking restart 2014-11-25 18:00:00 and get 2014-11-25 18:00:01 brctl: bridge br1: Resource busy cat: can't open '/sys/class/net/br1/brif/eth0/state': No such file or directory 2014-11-25 18:00:25 why do i get 'resource busy' even though i get a green ok ? 2014-11-25 18:02:47 br0 is ok 2014-11-25 18:03:07 and br0 and br1 have both bridge-ports eth0 2014-11-26 03:51:59 I'm not seeing the py-pip or py-virtualenv packages via apk search/info, but they do show up in the online package browser 2014-11-26 03:52:22 anyone know what might be the problem? 2014-11-26 03:53:30 http://dl-3.alpinelinux.org/alpine/v3.0/main is the only uncommented line in /etc/apk/repositories 2014-11-26 06:17:50 shodan45: it is in the v3.1 repository 2014-11-26 06:18:23 i think the online package browser tracks edge 2014-11-26 08:44:15 Good Morning all! 2014-11-26 08:48:07 <^7heo> moin ginjachris 2014-11-26 09:01:21 morning 2014-11-26 09:01:48 <^7heo> moin ScrumpyJack 2014-11-26 16:05:50 what's the "official" difference between main & testing package repos? 2014-11-26 16:06:23 I can't find any docs about testing 2014-11-26 16:08:53 I think testing is for new packages and such 2014-11-26 16:09:06 to make sure they run correctly before moving them to main 2014-11-26 16:10:02 testing is where new packages enters first 2014-11-26 16:10:21 once its tested that the new package builds and is okish, we move it to main 2014-11-26 16:10:50 the reason for the existance of 'testing' is that i am able to build packages for people 2014-11-26 16:11:04 but i dont have resources to test that all packages actually work 2014-11-26 16:11:32 so developers push it to 'testing' and let users test 2014-11-26 16:13:22 thanks 2014-11-26 16:22:35 what about testing vs. edge? 2014-11-26 17:32:11 <^7heo> shodan45: define "vs" 2014-11-26 17:36:22 differences 2014-11-26 21:48:51 hi 2014-11-26 21:48:58 i have an apk problem 2014-11-26 21:50:05 I have 2014-11-26 21:50:05 /etc/apk/cache -> /var/cache/apk 2014-11-26 21:51:00 and 2014-11-26 21:51:03 head -1 /etc/apk/repositories 2014-11-26 21:51:05 as 2014-11-26 21:51:10 /var/cache/apk 2014-11-26 21:52:00 when i run 2014-11-26 21:52:01 apk cache sync 2014-11-26 21:52:02 i get 2014-11-26 21:52:11 WARNING: Ignoring /var/cache/apk/x86_64/APKINDEX.tar.gz: No such file or directory 2014-11-26 21:52:17 I cannot understand why :( 2014-11-27 08:32:12 <^7heo_> moin 2014-11-27 08:54:48 just wondering, i know it's bloated but why not use a CMS like joomla for al.org? 2014-11-27 08:55:43 <^7heo> /kick ScrumpyJack 2014-11-27 08:56:18 :) 2014-11-27 08:56:21 <^7heo> /mode #alpine-linux +b ScrumpyJack$$nah_really,that_SUCKS 2014-11-27 08:56:28 <^7heo> ;) 2014-11-27 08:58:25 http://www.joomlart.com/demo/#purity_iii isn't too bad. but i guess it makes every website look the same 2014-11-27 08:58:47 http://purity_iii.demo.joomlart.com/ even 2014-11-27 08:59:02 <^7heo> ScrumpyJack: nah really 2014-11-27 08:59:05 <^7heo> please stop. 2014-11-27 08:59:10 <^7heo> that's not going to end well. 2014-11-27 08:59:14 <^7heo> 1. Web is shit 2014-11-27 08:59:30 <^7heo> 2. web processing languages are even MORE SHIT. 2014-11-27 09:00:02 <^7heo> 3. bloatd shitware made for managing the content over the 2 over the 1 is EVEN MORE BULLSHIT. 2014-11-27 09:00:07 <^7heo> so no, please stop. 2014-11-27 09:00:18 <^7heo> For the sake of the religion of the people who are here... 2014-11-27 09:00:19 <^7heo> please stop. 2014-11-27 09:00:50 i know the tech is not to everyone's taste, nor mine 2014-11-27 09:01:27 <^7heo> in a nutshell 2014-11-27 09:01:34 <^7heo> it's actually MORE work to do it with joomla. 2014-11-27 09:01:56 <^7heo> the ONLY reason why it takes so much time to come online is that we're a bit too perfectionists about it. 2014-11-27 09:02:22 <^7heo> it would take the sAME time with any shitty bloated CMS bullshit to be perfectionists, if not more. 2014-11-27 09:02:25 <^7heo> so that's not a valid solution. 2014-11-27 09:05:35 ncopa: hi, you created an docker image wich starts ssh by default. how can I add other services to be started when container ist started? 2014-11-27 09:07:03 xen_roger: it was not me who did docker 2014-11-27 09:07:11 iirc it was uggedal 2014-11-27 09:07:24 i have no idea on docker... 2014-11-27 09:07:50 ScrumpyJack: joomla? why not stick to drupal in that case? 2014-11-27 09:07:57 ohhh alpine-linux/base is tagged with your name as maintainer ;) 2014-11-27 09:08:04 ok ... ty anyway 2014-11-27 09:08:42 xen_roger: btw, i tweaked xendomains init script so it now supports running domUs in tmux 2014-11-27 09:09:02 so you can use tmux attach to get your domU consoles 2014-11-27 09:09:30 ncopa: sounds good ... also saw that openvswitch is now supported ;) 2014-11-27 09:09:38 yup 2014-11-27 09:09:49 i must say, i have underestimated xen 2014-11-27 09:09:57 its pretty neat 2014-11-27 09:10:08 <^7heo> Xen is kinda awesome 2014-11-27 09:10:12 <^7heo> especially after 4.2 2014-11-27 09:10:32 <^7heo> xl >>>>>>>>>>...>>>>>>>> xm 2014-11-27 09:10:47 best thing is that they have longtime support for sec fixes 2014-11-27 09:10:53 <^7heo> yeah 2014-11-27 09:10:54 which qemu/kvm does not... 2014-11-27 09:11:00 <^7heo> also 2014-11-27 09:11:06 <^7heo> it's xplaform 2014-11-27 09:11:10 <^7heo> which kvm isn't. 2014-11-27 09:11:18 <^7heo> and IOMMU is so great... 2014-11-27 09:35:22 Hello 2014-11-27 09:36:31 <^7heo> hello coredumb 2014-11-27 09:36:39 <^7heo> algitbot: I wasn't talking to you, idiot 2014-11-27 09:37:06 :D 2014-11-27 09:41:00 ncopa: wow, anything i can do to express thanks? 2014-11-27 09:41:34 o/ darkfader 2014-11-27 09:41:48 <^7heo> ship beers to alpine-linux headquarters, 42 /dev/null, 1337 Prøcessor, Sweden. 2014-11-27 09:42:08 <^7heo> ACTION hides 2014-11-27 09:53:59 would be nicer if the beers was sent to norway ;) 2014-11-27 09:55:34 <^7heo> ncopa: please provide full (non fictive) address then. 2014-11-27 09:56:44 darkfader: http://wiki.alpinelinux.org/wiki/Contribute 2014-11-27 09:57:44 <^7heo> ncopa: are you in oslo? 2014-11-27 09:58:05 ^7heo: 30 mins outside 2014-11-27 10:02:09 speaking of xen, thanks ncopa for removing perl/python in alpine-xen :) 2014-11-27 10:03:50 np 2014-11-27 10:10:51 setting up a new dom0: tmpfs 890.6M 62.2M 828.4M 7% / 2014-11-27 10:11:04 62.2MB 2014-11-27 10:11:04 <^7heo> ncopa: ok 2014-11-27 10:11:07 thats kinda nice :) 2014-11-27 10:11:18 i suppose you will want lvm and similar 2014-11-27 10:11:55 so i'd say a fully functional xen host is ~70MB 2014-11-27 10:12:20 which i must say is pretty nice 2014-11-27 10:13:22 openvswitch adds 8MB 2014-11-27 10:14:23 so, xen host with lvm, e2fsprogs, openvswitch, tmux is 81.1MB 2014-11-27 10:14:47 we want awall too 2014-11-27 10:14:54 85.4M 2014-11-27 10:28:51 <^7heo> is awall usable yet? 2014-11-27 10:29:58 yes 2014-11-27 10:30:06 its pretty nice 2014-11-27 10:30:47 the only thing i have mixed feelings about is json :) 2014-11-27 10:35:03 <^7heo> what would you like instead? 2014-11-27 10:35:04 <^7heo> yaml? 2014-11-27 10:43:41 yaml is easier for humans to edit 2014-11-27 10:43:53 i havent cared enough to write a patch for yaml 2014-11-27 10:43:57 looks relatively simple 2014-11-27 10:44:35 and its trivial to write a Makefile that converts the yaml to json files 2014-11-27 10:44:55 so i can write the rules in yaml, 'compile' to json so awall gets happy 2014-11-27 10:47:41 <^7heo> moin Jean-Scotch 2014-11-27 10:48:01 morning 2014-11-27 10:48:05 <^7heo> or you can directly implement yaml in awall 2014-11-27 10:48:07 <^7heo> that would work too. 2014-11-27 10:48:13 <^7heo> moin damhart_heinrich 2014-11-27 10:48:17 yo! 2014-11-27 10:48:28 morning ppl 2014-11-27 10:48:33 <^7heo> so yeah, I'd personally prefer to compile directly from yaml 2014-11-27 10:48:55 ^7heo: yes, implementing yaml in awall should be easy 2014-11-27 10:49:00 <^7heo> than yaml to json and then compile the rules.. 2014-11-27 10:50:29 <^7heo> wb damhart_heinrich 2014-11-27 10:51:38 :) 2014-11-27 11:03:46 loading 3.1 rc3 mini locally in virtualbox killed my X :) 2014-11-27 11:04:13 <^7heo> virtualbox did. 2014-11-27 11:04:16 <^7heo> ACTION hides 2014-11-27 11:04:35 sorry, alpine-xen, not alpine-mini 2014-11-27 11:05:41 <^7heo> ah yeah 2014-11-27 11:05:49 <^7heo> nah but that's another story entierely. 2014-11-27 11:06:04 <^7heo> you can't expect a virtualization system to give a sane env for another, totally different, one. 2014-11-27 11:06:24 <^7heo> especially something such as Xen, that is very low level compared to the other hypervisors. 2014-11-27 11:10:45 ScrumpyJack: xen should not work inside nested virt ... the other way around should work 2014-11-27 11:11:09 <^7heo> ^ 2014-11-27 11:11:26 <^7heo> Xen is really the best hypervisor around. 2014-11-27 11:11:31 <^7heo> nova aside, ofc. 2014-11-27 11:11:40 <^7heo> but nova isn't really mature. 2014-11-27 11:12:09 <^7heo> and nova didn't move since July, on the top of that. 2014-11-27 11:12:25 <^7heo> I guess that Herr Setinberg has better things to attend ;) 2014-11-27 11:13:40 <^7heo> I should try it btw. 2014-11-27 11:13:50 <^7heo> might be neat to have nova running on my laptop. 2014-11-27 11:13:56 never look at ... 2014-11-27 11:14:08 <^7heo> xen_roger: whatcha mean? 2014-11-27 11:14:29 < had never look at nova 2014-11-27 11:14:46 <^7heo> ah 2014-11-27 11:14:49 <^7heo> found it? 2014-11-27 11:14:51 <^7heo> or want the link? 2014-11-27 11:14:55 yes 2014-11-27 11:15:00 <^7heo> to which question? :D 2014-11-27 11:15:06 2nd 2014-11-27 11:15:08 ;) 2014-11-27 11:15:11 <^7heo> https://github.com/udosteinberg/NOVA 2014-11-27 11:15:26 ahh mean first :-P 2014-11-27 11:16:20 I still prefer stabe ( old ) stuff like xen ... expect docker ... docker is really nice 2014-11-27 11:17:09 <^7heo> xen_roger: stop with religion 2014-11-27 11:17:20 use it on my notebook to run some cisco stuff and other kind of 32bit sh** 2014-11-27 11:17:22 <^7heo> or state why you say that "xen is more stable" 2014-11-27 11:17:33 <^7heo> 'cause nova is "written" to be experimental 2014-11-27 11:17:36 not religion ... security ;) depends always on usecase 2014-11-27 11:17:43 <^7heo> but afaik, nobody here tested it. 2014-11-27 11:17:44 <^7heo> and 2014-11-27 11:17:46 <^7heo> security wise 2014-11-27 11:17:49 <^7heo> what you said is bs 2014-11-27 11:18:00 You can't expect security with virtualization 2014-11-27 11:18:00 <^7heo> the smaller the code base, the easier it is to audit 2014-11-27 11:18:11 <^7heo> Xen code base isn't exactly simple or small, to start with. 2014-11-27 11:18:31 <^7heo> on the other hand, you have the NOVA code base under your eyes :) 2014-11-27 11:18:37 xen has the most secure isolation as far as I know 2014-11-27 11:18:47 <^7heo> as far as you read, maybe. 2014-11-27 11:18:48 <^7heo> check it. 2014-11-27 11:18:51 <^7heo> Otherwise it's religion. 2014-11-27 11:18:57 <^7heo> Nova is a microkernel 2014-11-27 11:19:12 come on ... 2014-11-27 11:19:24 <^7heo> so it's basically doing linux + kvm in one block 2014-11-27 11:19:27 <^7heo> and doing THAT only. 2014-11-27 11:19:29 <^7heo> nah really, I mean it. 2014-11-27 11:19:36 <^7heo> I'm tired of ranting at various software without proof. 2014-11-27 11:19:42 <^7heo> so I'm not going to rant without proof anymore. 2014-11-27 11:19:53 <^7heo> afaik, hyper-v can totally be better than Xen. 2014-11-27 11:19:54 nobody is ranting ... 2014-11-27 11:20:05 <^7heo> ranting, debating, whatever. 2014-11-27 11:20:08 <^7heo> you got the point. 2014-11-27 11:20:14 <^7heo> discussing, in general. 2014-11-27 11:20:19 <^7heo> I'm a ranting guy, so I said ranting. 2014-11-27 11:20:22 <^7heo> but I meant discussing. 2014-11-27 11:20:29 ;) 2014-11-27 11:20:43 <^7heo> anyway 2014-11-27 11:20:47 <^7heo> I wanna try nova. 2014-11-27 11:20:48 <^7heo> Seems fun. 2014-11-27 11:20:52 <^7heo> and then I can compare with Xen. 2014-11-27 11:21:20 <^7heo> but it will likely take too much time to do a real comparison, detailed and in depth. 2014-11-27 11:21:21 feel free to report ... would be interesting. 2014-11-27 11:21:25 <^7heo> exactly ;) 2014-11-27 11:21:35 <^7heo> and then we can discuss based on what I/you/we find. 2014-11-27 11:21:45 <^7heo> and criticize the method, findings, improve. 2014-11-27 11:21:49 Oh, that's cpp 2014-11-27 11:22:02 time is always a point ... that´s the reason why I not read all the code and trust some guys how did 2014-11-27 11:22:11 xen is a microkernel 2014-11-27 11:22:41 <^7heo> royger: Xen is an hypervisor. 2014-11-27 11:22:57 <^7heo> USING a microkernel design. 2014-11-27 11:23:07 <^7heo> now, it's not able to run by itself, afaik. 2014-11-27 11:23:12 <^7heo> please prove me wrong if I am. 2014-11-27 11:23:22 <^7heo> but I believe that unfortunately I am not. 2014-11-27 11:23:33 <^7heo> it has to be tied to an existing kernel to work. 2014-11-27 11:23:46 well, it is able to run, but it just won't do anything useful without guests 2014-11-27 11:23:55 <^7heo> no it is not. 2014-11-27 11:24:03 <^7heo> you can't run it without an existing kernel. 2014-11-27 11:24:17 <^7heo> you need NetBSD's kernel, or Linux or whatever that Xen can run on. 2014-11-27 11:24:32 <^7heo> (i.e opensolaris) 2014-11-27 11:24:48 nope, Xen loads first, then Xen creates the Dom0 whatever (Linux, *BSD...) 2014-11-27 11:24:59 you can load Xen and nothing else, but it's just not useful 2014-11-27 11:25:31 you don't boot Linux and then boot Xen, it's the other way around 2014-11-27 11:25:46 <^7heo> okay, I admit that this is possible, but I would like to witness it. 2014-11-27 11:25:59 <^7heo> 'cause, afaik, it's tied to the kernel, not able to run by itself 2014-11-27 11:26:03 <^7heo> (i.e. missing parts) 2014-11-27 11:26:29 <^7heo> afaik, you boot a Xen-modified-Linux. 2014-11-27 11:26:46 <^7heo> which is to say that if you have Xen alone, you're missing parts (again). 2014-11-27 11:27:03 I agree with royger, there is also a console you could attach to it ... 2014-11-27 11:27:06 <^7heo> but I am not 100% certain, it's only what I understood after reading various doc and experimenting with it. 2014-11-27 11:27:24 <^7heo> xen_roger: I don't expect people to agree or disagree, we're not talking about opinions here. 2014-11-27 11:27:31 the mod-kernel is to let the dom0/domu know its running ontop of xen 2014-11-27 11:27:35 <^7heo> there are facts to be checked. 2014-11-27 11:27:56 <^7heo> Okay, I wanna see xen boot by itself, even if it panicks or do anything 2014-11-27 11:29:14 ^7heo: your kind of writing is some like agressive ... I don´t like that. 2014-11-27 11:29:18 <^7heo> I guess that the easiest way to do that is to check the Xen port of NetBSD. 2014-11-27 11:29:23 ever 2014-11-27 11:29:40 <^7heo> xen_roger: I told you many times that I don't like people bringing religion stuff to scientific questions 2014-11-27 11:29:50 <^7heo> and you "believe" and "agree" 2014-11-27 11:30:07 <^7heo> if I annoy you too much, you can /ignore me. You know that I guess. 2014-11-27 11:30:51 ^7heo: xen booting by itself: http://pastebin.com/raw.php?i=niyG1zUQ 2014-11-27 11:30:57 <^7heo> otherwise, please refrain to "agree" with people, you can however perfectly say that you have a proof that Xen can run without any dom0 2014-11-27 11:30:59 serial output, not much interesting 2014-11-27 11:31:01 <^7heo> then I'll check it out. 2014-11-27 11:31:11 <^7heo> royger: it is interesting, as it proves your point. 2014-11-27 11:31:21 <^7heo> royger: and therefore, if it is, teaches me more about Xen. 2014-11-27 11:31:24 <^7heo> in a way I can check. 2014-11-27 11:31:46 agreement != religion! anyway ... I was just confirming what royger said ... you act like someone how always know THE truth ... not really nice. 2014-11-27 11:32:29 <^7heo> okay, royger, you seem to be right. 2014-11-27 11:32:41 <^7heo> I am sorry I doubted what you said, but I don't believe strangers on irc ;) 2014-11-27 11:32:46 also nova codebase is much more smaller because it requires VTx/AMD-V in order to work, Xen still has old the previous PV stuff from times before x86 had hardware virtualization 2014-11-27 11:32:53 <^7heo> however, I'll check that myself later. 2014-11-27 11:32:58 <^7heo> it is interesting. 2014-11-27 11:33:20 <^7heo> then I understand also much better why the NetBSD Xen port is based on x86 2014-11-27 11:35:53 xen doesn't introduce a new instruction set, it only virtualizes either x86 or ARM, but then you need to use hypercalls for some operations 2014-11-27 11:36:18 <^7heo> ok 2014-11-27 11:36:28 <^7heo> so in theory 2014-11-27 11:36:33 <^7heo> you can do 3 NetBSD Xen ports 2014-11-27 11:36:39 <^7heo> x86 x86-64 and ARM 2014-11-27 11:38:19 <^7heo> anyhow, that's not the topic. But thanks for the info. 2014-11-27 11:38:20 well, it's basically a matter of picking the bare metal version and then add hooks to some operations that cannot be performed when running as a Xen guest 2014-11-27 11:38:30 <^7heo> yeah I see. 2014-11-27 11:42:15 <^7heo> so the only diff between Xen and Nova is that Nova implements the Dom0 too 2014-11-27 11:42:24 <^7heo> allowing you to focus on guests 2014-11-27 11:42:44 <^7heo> and implements only hardware-based virt too 2014-11-27 11:45:10 I guess so, also device drivers seem to be inside of the hypervisor itself, IMHO this is not a great idea 2014-11-27 11:45:41 <^7heo> yeah but if you implement the Dom0, what other choices have you got? 2014-11-27 11:45:51 <^7heo> unless you rely on IOMMU too... 2014-11-27 11:45:55 <^7heo> but that's a hell of a requirement. 2014-11-27 11:47:59 that's where Xen is headed, PVH Dom0 requires an IOMMU (classic PV of course doesn't) 2014-11-27 11:48:38 <^7heo> really? 2014-11-27 11:48:46 <^7heo> since when/what version? 2014-11-27 11:49:15 4.5 will be the first version with PVH Dom0 support 2014-11-27 11:49:39 <^7heo> ok 2014-11-27 12:18:13 royger, Xen does not require VTx/AMD-V in order to work ? 2014-11-27 12:18:22 or utilizes it 2014-11-27 12:22:32 vkrishn: XEn requires VTx/AMD-V for HVM guests, if you only run PV guests you don't need hw virt extensions 2014-11-27 12:22:48 ok, thanks 2014-11-27 12:41:30 <^7heo> vkrishn: you can do without hardware virt if you do a paravirtualized machine. 2014-11-27 12:41:37 <^7heo> but for that you need a modified guest. 2014-11-27 12:41:43 <^7heo> so, windows not possible. 2014-11-27 12:42:04 <^7heo> and that is pretty awesome. 2014-11-27 15:10:31 new website launched 2014-11-27 15:12:59 \o/ 2014-11-27 15:13:33 Great ! 2014-11-27 15:15:40 Is the width of the Latest Development dates intended to cut the year from the rest ? 2014-11-27 15:16:09 Hum, that's just on my browser 2014-11-27 15:16:51 On chromium too 2014-11-27 15:21:58 quinq: screenshot? 2014-11-27 15:21:58 <^7heo> ncopa: \o/ 2014-11-27 15:22:09 ^7heo: big thanks for the help! 2014-11-27 15:22:26 <^7heo> ncopa: oh, I wanted to submit a last patch 2014-11-27 15:22:33 <^7heo> the logo should be clickable imho 2014-11-27 15:22:38 <^7heo> and should link at the same place at home. 2014-11-27 15:22:43 would be nice indeed 2014-11-27 15:22:46 <^7heo> I do click on logos ALL the time. 2014-11-27 15:22:48 i even thought of removing 'home' 2014-11-27 15:22:55 <^7heo> yeah, would make sense. 2014-11-27 15:23:00 and let user click on the logo for home 2014-11-27 15:23:09 <^7heo> yeah 2014-11-27 15:23:14 making logo clickable is first step though 2014-11-27 15:23:16 <^7heo> Well it doesn't hurt to leave home 2014-11-27 15:23:22 yah 2014-11-27 15:23:31 <^7heo> so people used to the navigation would still feel at "home" :D 2014-11-27 15:23:36 <^7heo> ACTION hides 2014-11-27 15:24:16 clandmeter: drupal site is still available via https. but i think that reousrce to css'es are blocked because they are hardcoded to http:// 2014-11-27 15:24:25 and https does not like mixing with http 2014-11-27 15:24:30 <^7heo> right 2014-11-27 15:24:48 clandmeter: any idea how to fix that in drupal? 2014-11-27 15:24:52 Ok, just a minute 2014-11-27 15:25:28 ncopa: its because of the rev proxy 2014-11-27 15:25:43 i dont have time right now. so i cannot check 2014-11-27 15:25:48 ok... 2014-11-27 15:25:55 i'll try find a hack for now 2014-11-27 15:28:24 <^7heo> so wait, you have a reverse proxy that takes the http website and serves it over https, or? 2014-11-27 15:30:16 http://s18.postimg.org/tw4geyq2x/al_ffox.png 2014-11-27 15:30:17 http://s16.postimg.org/o2hkzr2px/al_chromium.png 2014-11-27 15:30:21 http://s12.postimg.org/aoif1ms6l/al_wk2gtk.png 2014-11-27 15:31:09 ^7heo: opposite. we have a front nginx server 2014-11-27 15:31:15 that takes http/https 2014-11-27 15:31:25 and forwards to different backends over httå 2014-11-27 15:31:27 http* 2014-11-27 15:31:39 <^7heo> ah 2014-11-27 15:31:46 the reverse proxy and backends runs on same iron 2014-11-27 15:31:52 <^7heo> oky 2014-11-27 15:31:54 <^7heo> okay 2014-11-27 15:31:55 even same vm 2014-11-27 15:32:02 http://s23.postimg.org/k6maw2wor/al_lynx.png 2014-11-27 15:32:05 <^7heo> and why a reverse then? 2014-11-27 15:32:34 because we reuse the same public ip for various hosts 2014-11-27 15:32:47 the hosts runs in different vservers 2014-11-27 15:33:40 quinq: i saw that on my tablet 2014-11-27 15:33:47 havent taken time to fix it... 2014-11-27 15:34:07 <^7heo> ah okay 2014-11-28 12:40:04 <^7heo> say 2014-11-28 12:40:27 <^7heo> if I have an "unsatisfiable constraint" with apk 2014-11-28 12:40:34 <^7heo> after apk update 2014-11-28 12:40:50 <^7heo> how can I solve it? 2014-11-28 12:40:52 <^7heo> what does it mean? 2014-11-28 12:48:43 it means that apk is not able to figure out the dependencies 2014-11-28 12:49:26 might be that some dependency needs something that conflics with what other dependency needs 2014-11-28 12:50:47 <^7heo> hmm 2014-11-28 12:50:54 <^7heo> how can I get more info about it? 2014-11-28 13:27:16 ^7heo: apk dot --errors 2014-11-28 13:27:53 <^7heo> thanks 2014-11-28 13:28:26 <^7heo> ncopa: is that the complete log of ALL the errors I ever got with APK? 2014-11-28 13:28:42 thats a log of dependency errors in repo 2014-11-28 13:28:53 <^7heo> ah, so not only in the installed base 2014-11-28 13:29:06 <^7heo> mplayer isn't found in that list. 2014-11-28 13:33:51 <^7heo> ncopa: I have more information, I should maybe have told you before. mplayer was installed with @stable pinning 2014-11-28 13:33:57 <^7heo> and I wanted to install the current one. 2014-11-28 13:33:58 <^7heo> to test it 2014-11-28 13:34:12 <^7heo> so I did apk del mplayer && apk add mplayer 2014-11-28 13:34:28 <^7heo> oh, fun. 2014-11-28 13:34:33 <^7heo> when I install it again at stable... 2014-11-28 13:34:48 <^7heo> apk "downgrades" the two libs that weren't found... 2014-11-28 13:35:25 <^7heo> okay, it's a problem with upgrading/downgrading libs. 2014-11-28 13:35:50 <^7heo> can anyone with edge please try to install mplayer? 2014-11-28 13:35:55 <^7heo> (w/out pinning) 2014-11-28 13:37:18 <^7heo> okay nevermind, I found the problem I guess. 2014-11-28 13:37:31 <^7heo> ncopa: dependency mismatch on mplayer@edge as far as I can tell 2014-11-28 13:37:57 <^7heo> the point is, on my computer, I'm tracking edge, and pinning @stable and @testing 2014-11-28 13:38:29 <^7heo> if I "apk add mplayer" I get an error for two libs that aren't possibly satisfiable 2014-11-28 13:38:43 <^7heo> but if I install them both @stable PRIOR to installing mplayer, then all works. 2014-11-28 13:39:07 <^7heo> so I guess that mplayer@edge has the wrong dependency constraints. 2014-11-28 13:52:20 I think I had this issue with mpv 2014-11-28 13:52:47 Or webkitgtk2 2014-11-28 14:14:14 ^7heo: should work now 2014-11-28 14:14:53 <^7heo> ncopa: already confirmed on -devel 2014-11-28 20:00:21 http://www.alpinelinux.org/packages/ doesn't work 2014-11-28 20:01:04 on chromium 2014-11-28 20:01:44 but it does on firefox 2014-11-28 20:02:43 413 Request Entity Too Large, Your request was dropped because it was too long. 2014-11-28 20:03:53 also, all files give a page not found 2014-11-28 21:24:17 works for me on chromium 2014-11-28 21:28:13 same here 2014-11-28 21:28:30 (but links from packages or indeed empty) 2014-11-29 15:47:45 Saludos, alguien que hable español ? 2014-11-29 20:56:42 umm 2014-11-29 20:56:51 how can i install a package from testing repo? 2014-11-29 21:00:14 ah 2014-11-29 21:00:36 i was trying @testing but actually it is @edge/testing 2014-11-30 01:41:13 Is it possible to get notified automatically (like with aptitron on Debian) about new packages pending an upgrade? 2014-11-30 01:41:44 aptitron->apticron 2014-11-30 01:43:39 Notified by email that is. 2014-11-30 03:01:40 ncopa, new site looks really messed up on android 2014-11-30 11:44:20 <^7heo> dalias: who care about android? 2014-11-30 15:00:45 re apticron: I use a zabbix trigger that alerts me as soon as `apk --simulate upgrade|head -n1|sed 's/^OK: /(0\/0)/;s/.\d\/\(\d\).*/\1/'` becomes greater 0