#archlinux32 | Logs for 2021-05-28

Back
[00:47:55] -!- drathir_tor has quit [Ping timeout: 252 seconds]
[01:06:42] <KillerWasp> hello, qemu lie in something side about of the files list, if i use 'pkgfile -s/-l' show /usr/bin/qemu-system-i386, but from 'pacman -S/-Ql' don't exist qemu-system-i386 file.
[01:08:05] <KillerWasp> i guess pkgfile database are based in archlinux 64bit system.
[01:12:06] <KillerWasp> If I'm not wrong about the processor, I assume I'm using this package: http://archlinux32.org - Although 'uname -a' shows i686.
[01:12:09] <phrik> Title: Arch Linux 32 - qemu 5.2.0-4.2 (pentium4) (at archlinux32.org)
[01:14:58] -!- drathir_tor has joined #archlinux32
[01:21:54] <KillerWasp> can't show the maintainer list? Looks like it was updated 1 month ago.
[01:49:23] <KillerWasp> ok, I have already realized that it is exactly the same as Archlinux 64 bit, without any difference.
[01:49:23] -!- doskoi has quit [Quit: kthxbye]
[01:50:29] -!- doskoi has joined #archlinux32
[02:05:04] -!- drathir_tor has quit [Changing host]
[02:05:04] -!- drathir_tor has joined #archlinux32
[06:09:44] -!- slacka has quit [Quit: Leaving]
[06:50:11] <girls> KillerWasp: qemu-system-i386 (which is for all 32-bit guests) is part of qemu-arch-extra
[06:50:26] <girls> buildmaster: wtf qemu-system-i386
[06:50:34] <buildmaster> girls: qemu [extra], qemu-headless [extra], qemu [staging], qemu-headless [staging], qemu [testing], qemu-headless [testing]: /usr/bin/qemu-system-i386
[06:50:34] <buildmaster> bash-completion [extra]: /usr/share/bash-completion/completions/qemu-system-i386
[06:51:16] <girls> huh? well, pacman -Qo on my arch system said so :-/
[06:57:24] -!- titus_livius has joined #archlinux32
[06:59:38] -!- girls has quit [Quit: ZNC 1.8.2 - https://znc.in]
[07:00:22] -!- girls has joined #archlinux32
[07:00:23] <buildmaster> Hi girls!
[07:00:23] <buildmaster> !rq girls
[07:00:24] <phrik> buildmaster: <girls> an inefficient slave is better than no slave at all
[07:09:53] -!- eschwartz has quit [Ping timeout: 252 seconds]
[07:18:25] <KillerWasp> no, not is qemu-i386.
[07:23:47] -!- eschwartz has joined #archlinux32
[07:29:02] -!- eschwartz has quit [Ping timeout: 248 seconds]
[07:31:46] -!- eschwartz has joined #archlinux32
[07:52:14] -!- eschwartz has quit [Ping timeout: 252 seconds]
[08:12:06] -!- eschwartz has joined #archlinux32
[09:02:05] <girls> ah, I see the problem: /usr/bin/qemu-system-x86_64 is part of qemu on archlinux, but not on archlinux32 (obviously) - however, qemu-system-i386 *also* is not part of qemu on archlinux32, but rather part of qemu-arch-extra - like on archlinux (where this is correct)
[09:05:44] <girls> this is a pentium4 problem: pentium4 is not correctly spotted by the PKGBUILD
[09:05:49] <girls> i686 is fine
[09:06:00] <girls> and i486 also
[09:12:38] -!- City-busz has joined #archlinux32
[09:39:13] -!- abaumann has joined #archlinux32
[09:39:13] <buildmaster> Hi abaumann!
[09:39:13] <buildmaster> !rq abaumann
[09:39:14] <phrik> buildmaster: <abaumann> classic approach then.. add more resources to the problem :-)
[09:39:28] <girls> Hi abaumann!
[09:40:39] <abaumann> morning :-)
[09:46:55] <girls> btw, abaumann: I think, you're op in #archlinux-ports, now, too (at least, I asked anthraxx for it :-D)
[09:47:07] <abaumann> uh. :-)
[13:30:22] -!- buildmaster has quit [Remote host closed the connection]
[13:32:48] -!- buildmaster has joined #archlinux32
[13:32:48] <buildmaster> !rq buildmaster
[13:32:49] <phrik> buildmaster: <buildmaster> I might be insane, but never confused ... ;-)
[14:28:04] -!- eschwartz has quit [Quit: A random quit message]
[14:32:02] -!- abaumann has quit [Read error: Connection reset by peer]
[14:35:02] -!- eschwartz has joined #archlinux32
[15:01:53] -!- abaumann has joined #archlinux32
[15:01:53] <buildmaster> Hi abaumann!
[15:01:53] <buildmaster> !rq abaumann
[15:01:54] <phrik> buildmaster: <abaumann> I guess it's the usual bug: transactions are not transactions on mysql
[17:25:16] -!- slacka has joined #archlinux32
[17:31:10] -!- abaumann has quit [Quit: leaving]
[17:32:53] -!- slacka has quit [Remote host closed the connection]
[17:40:53] -!- slacka has joined #archlinux32
[17:52:46] <slacka> T`aZ: here is a diff that created error: redefinition of 'DebianMultiarch'
[17:53:50] <slacka> https://pastebin.com
[17:53:51] <phrik> Title: // Try /../$triple/include/c++/$version then /../include/c++/$version. if ( - Pastebin.com (at pastebin.com)
[17:55:06] <slacka> https://pastebin.com
[17:55:07] <phrik> Title: code in context - Pastebin.com (at pastebin.com)
[17:56:00] <slacka> the error makes sense, but removing the 2nd def, doesn't fix the original problem
[17:57:12] <T`aZ> that's not a diff
[17:57:30] <T`aZ> how come you added StringRef DebianMultiarch into the argument of addGCCLibStdCxxIncludePaths
[18:01:08] <T`aZ> this is why it fails btw, you have both an argument called DebianMultiarch , and a variable a bit later with the same name
[18:01:56] <T`aZ> so just remove StringRef DebianMultiarch from the arguments of addGCCLibStdCxxIncludePaths
[18:27:38] -!- fisch02 has joined #archlinux32
[19:10:14] <slacka> T`aZ: I didn't add it, that was from your pastebin, but when i did the build still fails on Arch 32
[19:12:02] <slacka> that's what I meant, by removing the 2nd definition, doesn't fix the original problem
[19:12:47] <T`aZ> the error is with the parameter, it should not be there
[19:13:02] <T`aZ> i mean, based on the original patch linked to the issue
[19:13:16] <slacka> oh in the bugtracker?
[19:13:19] <slacka> ok
[19:13:19] <T`aZ> yes
[19:14:06] <slacka> I agree the best solution is to have both code paths
[19:14:35] <slacka> what I'm not sure of is on the logic to decide which one
[19:15:30] <T`aZ> my understanding is that it tries a bunch of stuff until it works, so if you try to add something non existant, it will not matter
[19:15:50] <T`aZ> but im not familiar with that project so
[19:21:25] <slacka> yeah you just try to keep running it with your patch, right?
[19:21:59] <slacka> repeat addLibStdCXXIncludePaths()
[19:22:24] <slacka> I'm not familiar with this code either, nor c/cpp
[19:32:05] <slacka> T`aZ: so including both seems to be working for arch 32
[19:32:06] <slacka> https://pastebin.com
[19:32:08] <phrik> Title: diff of both include calls - Pastebin.com (at pastebin.com)
[19:33:08] <T`aZ> nice to hear
[19:34:10] <T`aZ> but if you post this patch, please move the original comment // Detect Debian g++-multiarch-incdir.diff just below the added line, because it is for the debianmultiarch part :p
[19:55:37] <slacka> T`aZ: that was just a test. While the build works, it breaks several unit tests that check include paths
[19:56:47] <slacka> unit test failure shows the include path has both the debian (i386) and the arch (i686) path
[19:58:47] <T`aZ> ha, you might want to get in contact with the maitainer of those tests
[20:01:00] <slacka> running it on opensuse i686 to compare
[20:03:37] <T`aZ> it all boils down to : why is the path different, is somebody in the wrong ? if nobody in the wrong, handling those 2 cases should be supported, but then maybe the unit tests are not correct (this needs to be confirmed)
[20:20:33] <slacka> T`aZ: I think the issue is this code:
[20:20:34] <slacka> StringRef DebianMultiarch =
[20:20:34] <slacka> GCCInstallation.getTriple().getArch() == llvm::Triple::x86
[20:21:24] <slacka> won't all 32-bit distros == llvm::Triple::x86
[20:21:40] <T`aZ> i dont know what llvm::Triple::x86 boild down to
[20:21:47] <slacka> and if so, that makes use the incorrect Debian path, rather than the default that works
[20:21:56] <slacka> neither do I
[20:22:13] <slacka> but that code path breaks Arch 32 and if I force it back, it works
[20:24:11] <slacka> https://llvm.org
[20:24:14] <phrik> Title: LLVM: llvm::Triple Class Reference (at llvm.org)
[20:24:35] <slacka> correct me if I'm wrong, but I read that as If the architecture is x86, use the Debian specific code
[20:24:56] <slacka> so Arch 32 and OpenSuse i686 break
[20:27:42] <T`aZ> so, if indeed it is x86, it will try i386-linux-gnu
[20:27:46] <T`aZ> in all cases
[20:28:13] <T`aZ> otherwise, the generic (nonx86) triple
[20:29:07] <T`aZ> so, why did they chose i386 instead of the generic triple, who is in the wrong ?
[20:29:31] <T`aZ> what is archlinux32 using for the i486 case btw ?
[20:29:59] <slacka> Hmmm, I'm going to try to remove the ?:, just set StringRef DebianMultiarch = TripleStr;
[20:30:09] <slacka> see what happens
[20:32:49] <T`aZ> that will break builds on debian :D
[20:35:17] <slacka> well they just broke all builds on OpenSUSE / Arch32...seems fair :)
[20:35:47] <slacka> I just want to see if that still breaks the 2 tests the fail with my last attempt
[20:36:01] <slacka> just shooting in the dark
[20:36:25] <slacka> if I can get a reasonable patch to submit maybe someone that understand the code can help
[20:39:20] <slacka> I have to run, if you think of anything else to try or have any other suggestions, please leave a not on the bugtracker if I'm not here
[20:39:34] <slacka> leave a note\
[20:41:44] <T`aZ> to be honest i have no other quick ideas, I would need to get more involved, and a) no time, b) not interested :o
[20:42:12] <T`aZ> anything acknowledging distro specific behaviour should die in a fire
[20:42:30] <T`aZ> so that does not help :p
[20:43:05] <slacka> Well Debian is the outlier here. But good luck getting them to change their ways
[21:13:39] <slacka> StringRef DebianMultiarch = TripleStr;
[21:13:45] <slacka> so far has fixed everything
[21:21:13] <T`aZ> that is equivalent to the original code, without the debian patch
[21:21:23] <T`aZ> iirc
[22:59:04] -!- petris_ has joined #archlinux32
[22:59:04] -!- petris has quit [Ping timeout: 264 seconds]
[23:15:48] -!- fisch02_ has joined #archlinux32
[23:18:52] -!- fisch02 has quit [Ping timeout: 264 seconds]
[23:18:54] fisch02_ is now known as fisch02
[23:29:32] -!- slacka has quit [Quit: Leaving]
[23:40:30] -!- KillerWasp has quit [Ping timeout: 264 seconds]
[23:45:19] -!- fisch02 has quit [Quit: fisch02]