#archlinux32 | Logs for 2018-12-10

Back
[00:15:13] -!- guys has quit [Ping timeout: 252 seconds]
[00:16:02] -!- guys has joined #archlinux32
[00:47:20] <elibrokeit> deep42thought, abaumann: I tried to replace glibc with musl in pacman-static, by adding makedepends on musl and kernel-headers-musl, then export CC=musl-gcc CXX=musl-gcc
[00:47:27] <elibrokeit> this works fine on x86_64
[00:48:04] <elibrokeit> on i686 it crashed while building libcares:
[00:48:04] <elibrokeit> /usr/bin/ld: ahost-ahost.o: in function `callback':
[00:48:04] <elibrokeit> ahost.c:(.text+0x167): undefined reference to `__stack_chk_fail_local'
[01:02:13] -!- rcf has quit [Read error: Connection reset by peer]
[01:17:36] -!- rcf has joined #archlinux32
[01:19:27] -!- zro has quit [Ping timeout: 240 seconds]
[01:21:41] -!- zro has joined #archlinux32
[02:57:08] -!- woshty has quit [Ping timeout: 245 seconds]
[04:07:51] -!- guys has quit [Ping timeout: 252 seconds]
[04:11:40] -!- guys has joined #archlinux32
[05:05:54] -!- oaken-source has joined #archlinux32
[05:48:13] -!- oaken-source has quit [Quit: leaving]
[07:39:57] -!- NoobAlice has quit [Quit: Leaving.]
[07:53:30] -!- djmoch has quit [Ping timeout: 246 seconds]
[08:58:54] -!- deep42thought has joined #archlinux32
[08:58:54] <buildmaster> Hi deep42thought!
[08:58:54] <buildmaster> !rq deep42thought
[08:58:55] <phrik> buildmaster: <deep42thought> ah, maybe, borg is borged, too
[09:21:45] -!- djmoch has joined #archlinux32
[10:39:27] -!- guys has quit [Ping timeout: 252 seconds]
[12:17:24] -!- woshty has joined #archlinux32
[12:20:53] <buildmaster> i686/purpose is broken (says nlopc46).
[12:30:42] <buildmaster> i686/extra-cmake-modules are broken (says nlopc46).
[12:35:02] <buildmaster> i686/breeze-icons are broken (says nlopc46).
[12:38:31] <buildmaster> i686/oxygen-icons are broken (says nlopc46).
[13:25:05] <buildmaster> i686/linux is broken (says eurobuild3).
[13:29:22] <buildmaster> i686/linux-zen is broken (says rechenknecht).
[13:35:04] -!- lodger has joined #archlinux32
[13:36:41] -!- deep42thought has quit [Remote host closed the connection]
[13:40:17] -!- deep42thought has joined #archlinux32
[13:40:17] <buildmaster> Hi deep42thought!
[13:40:17] <buildmaster> !rq deep42thought
[13:40:19] <phrik> buildmaster: <deep42thought> brtln: you're "the arch linux", we're "the other arch linux" and alarm is "arches linux", then?
[13:52:52] -!- lodger has quit [Remote host closed the connection]
[14:02:27] -!- woshty has quit [Ping timeout: 240 seconds]
[14:22:08] <buildmaster> i686/linux is broken (says rechenknecht).
[14:27:34] <buildmaster> i686/linux-zen is broken (says rechenknecht).
[14:33:06] <buildmaster> i686/buildbot-www is broken (says rechenknecht).
[15:44:10] -!- bill-auger has quit [Ping timeout: 250 seconds]
[15:45:00] -!- bill-auger has joined #archlinux32
[16:15:01] -!- deep42thought has quit [Quit: Leaving.]
[16:20:16] -!- guys has joined #archlinux32
[16:57:14] -!- woshty has joined #archlinux32
[17:35:01] <buildmaster> i686/rustup is broken (says rechenknecht).
[17:35:47] -!- bill-auger has quit [Ping timeout: 240 seconds]
[17:37:20] -!- bill-auger has joined #archlinux32
[17:55:07] -!- bill-auger has quit [Ping timeout: 240 seconds]
[17:56:12] -!- bill-auger has joined #archlinux32
[18:20:52] -!- MrBIOS has quit [Quit: MrBIOS]
[18:29:27] -!- woshty has quit [Read error: Connection reset by peer]
[18:29:49] -!- woshty has joined #archlinux32
[18:55:02] -!- MrBIOS has joined #archlinux32
[19:13:01] -!- abaumann has joined #archlinux32
[19:13:01] <buildmaster> Hi abaumann!
[19:13:05] <buildmaster> !rq abaumann
[19:13:05] <phrik> buildmaster: <abaumann> YASMSBOP: yet another silly make system based on python
[19:14:58] <abaumann> elibrokeit: I encountered that while building a standalone C-based kernel, somehow gcc generates some calls to a local stack smashing function, which is then never defined, I solved it by hacking my own one, see http://git.andreasbaumann.cc
[19:15:00] <phrik> Title: kernel.c\kernel\src - abaos - [no description] (at git.andreasbaumann.cc)
[19:16:14] <abaumann> Now for pacman-static this is hardly a solution, there must be a library somewhere which contains implementations of those stack smashing functions, they don't do much more than merilly printing an error message like 'stack smashing detected' and maybe playing a little bit with libunwind to present the user a stack.
[19:18:10] <abaumann> There is a -lssp_nonshared LDFLAGS, but I don't really know where this library comes from.
[19:18:44] <abaumann> *abaumann tries to build by adding -lssp_nonshared to LDFLAGS..
[19:19:52] <elibrokeit> Well, there must be some way to build C programs on i686, so this is something in need of a solution
[19:20:19] <elibrokeit> hopefully this solution is not "give up on -fstack-protector=strong for i686"
[19:20:20] <abaumann> The problem is -fstack-protector-all, I guess.
[19:20:24] <abaumann> :-)
[19:20:38] <elibrokeit> but if so, then makepkg.conf has some... issues.
[19:20:54] <elibrokeit> note: it works fine for glibc
[19:20:58] <elibrokeit> not for musl libc
[19:21:15] <elibrokeit> *supposedly* fixed in http://git.musl-libc.org
[19:21:16] <phrik> Title: musl - musl - an implementation of the standard library for Linux-based systems (at git.musl-libc.org)
[19:21:18] <abaumann> I remember it was an issue in musl discussed about..
[19:22:27] <abaumann> mmh. adding it brutally to LDFLAGS as some libraries don't like libssp_nonshared.a
[19:24:08] <elibrokeit> ar x /usr/lib/musl/lib/libc.a __stack_chk_fail.lo
[19:24:21] <elibrokeit> strings __stack_chk_fail.lo
[19:24:22] <abaumann> aha. it was a separate library before?
[19:24:26] <elibrokeit> >> __stack_chk_fail_local
[19:24:43] <elibrokeit> y u no worky
[19:32:17] <elibrokeit> abaumann: if you know why musl-gcc does *not* find this symbol on i686, but does on x86_64, I'm all ears.
[19:32:40] <elibrokeit> I'd like musl binaries of pacman-static, because glibc is not "really" static due to libnss_*.so
[19:32:43] <abaumann> yep. I'm digging. Just my machine is not the fastest one.. :-)
[19:45:56] -!- deep42thought has joined #archlinux32
[19:45:57] <buildmaster> Hi deep42thought!
[19:45:57] <buildmaster> !rq deep42thought
[19:45:58] <phrik> buildmaster: <deep42thought> so you're a garbage collector?
[19:46:30] <abaumann> hi deep42thought
[19:46:35] <deep42thought> Hi abaumann!
[19:48:10] <deep42thought> !bug 61018
[19:48:11] <phrik> https://bugs.archlinux.org
[19:48:12] <deep42thought> oops :-D
[19:48:27] * deep42thought is too pedantic
[19:49:52] <abaumann> yeah, maybe that's exagerated. :-)
[19:50:21] <deep42thought> I just noticed one of my packages compiled from aur failed due to broken linking
[19:50:32] <deep42thought> and I thought: hey, makepkg has a feature for that
[19:55:42] <elibrokeit> deep42thought: you would need to modify PKGBUILDs to link depend on libidns.so too, so it wouldn't help you handle large distro staging builds
[19:55:56] <deep42thought> no, not for archlinux32
[19:56:03] <deep42thought> but for packages compiled from aur it would help
[19:56:06] <elibrokeit> s/link//; s/libidns.so/libidn.so/
[19:56:17] <deep42thought> e.g. I would see pacman failing to update libidn instead of software failing to run
[19:56:34] <elibrokeit> Packages compiled from AUR are the worst, because then you cannot pacman -Syu and update the AUR package after
[19:57:00] <elibrokeit> you need to uninstall the software and update. Or exclusively use clean chroot builds.
[19:57:01] <deep42thought> not if you compile the aur packages in a clean chroot
[19:57:16] <deep42thought> I usually build aur packages on a separate host in a clean chroot
[19:57:18] <elibrokeit> Yes, and how many AUR users do that
[19:57:45] <deep42thought> there should be aur-x86_64-build ;-P
[19:57:59] <deep42thought> alongside extra-x86_64-build and co.
[19:58:02] <elibrokeit> It's something I dislike about voidlinux, which adds such soname deps automatically and assumes all packages are built by the official repos
[19:58:17] <elibrokeit> deep42thought: it's called "aurutils" => "aur build"
[19:58:39] <deep42thought> whoah, that exists?
[19:58:43] * deep42thought searches
[19:59:17] <elibrokeit> aurutils is best AUR helper
[19:59:28] <elibrokeit> the only one with native, first-class support for clean chroots
[20:03:28] <deep42thought> yeah, the concept sounds clean and sane
[20:08:51] -!- davor has quit [Ping timeout: 246 seconds]
[20:09:58] -!- davor has joined #archlinux32
[20:20:27] <deep42thought> elibrokeit: so why not add provides entries for all/major libraries?
[20:20:49] <elibrokeit> deep42thought: why yes?
[20:21:04] <deep42thought> because I won't accumulate broken software
[20:21:12] <elibrokeit> we assume people don't use partial updates and typically do not cater to versioned dependencies
[20:21:21] <deep42thought> how do I get notified, that I need to recompile certain packages?
[20:21:27] <elibrokeit> Devtools has lddd to find broken packages
[20:21:42] <elibrokeit> it will find all broken binaries or libs that are currently installed
[20:22:09] <deep42thought> after it broke
[20:23:11] <deep42thought> may I ask, why that makepkg feature (of tracking library versions) exists if it's not supposed to be used?
[20:24:02] <deep42thought> I know, that's like this discussion why versioned dependencies exist in the first place ...
[20:26:10] <elibrokeit> right :)
[20:26:26] <elibrokeit> also, because makepkg/pacman is not exclusive to Arch Linux
[20:26:56] <elibrokeit> e.g. you use it too ;) so does MSYS2 on Windows, so do several other linux distros -- a couple of which aren't even based on Arch
[20:27:16] <deep42thought> uh, yeah, right
[20:27:17] <elibrokeit> Chakra has completely forked the package repositories and now maintains all packages in-house
[20:27:21] <deep42thought> I constantly forget about that
[20:27:53] <elibrokeit> I believe they do the "stability" thing, so quite possibly they depend heavily on soname deps.
[20:28:33] <elibrokeit> Frugalware used to be pacman contributors, but not for years now, and their fork of pacman/makepkg is now significantly diverged.
[20:28:52] <elibrokeit> Or actually maybe not as I don't know how heavily they develop it anymore ;)
[20:29:09] <deep42thought> :-p
[20:29:48] <deep42thought> I'm sometimes fooled by the "the feature is there - you must use it" fallacy
[20:42:40] <abaumann> seabios: /usr/share/qemu/vgabios-virtio.bin exists in filesystem (owned by qemu)
[20:42:47] <abaumann> mmh. a clash with qemu..
[20:43:06] <abaumann> we apply force (better known as overwrite nowadays) :-)
[20:43:43] <elibrokeit> https://bugs.archlinux.org
[20:43:44] <phrik> Title: FS#60726 : [seabios] File conflict in seabios and qemu (at bugs.archlinux.org)
[20:44:20] <elibrokeit> (which is *not* a duplicate of https://bugs.archlinux.org due to it repeatedly happening)
[20:44:21] <phrik> Title: FS#60723 : [seabios] Unable to upgrade to seabios-1.11.0-2 (at bugs.archlinux.org)
[20:44:58] <abaumann> I could argue that this is what comments have been invented for :-)
[20:45:42] <abaumann> mmh, no luck with LD=musl-gcc
[21:02:59] -!- abaumann has quit [Quit: leaving]
[23:12:58] -!- NoobAlice has joined #archlinux32
[23:23:13] -!- deep42thought has quit [Quit: Leaving.]
[23:26:58] -!- guys has quit [Ping timeout: 250 seconds]