#archlinux32 | Logs for 2019-05-15
Back
[00:33:34] -!- ofara_ has quit [Ping timeout: 255 seconds]
[01:08:23] -!- guys has quit [Ping timeout: 258 seconds]
[01:46:03] -!- ofara_ has joined #archlinux32
[01:47:34] -!- guys has joined #archlinux32
[01:50:00] -!- MrBIOS has quit [Quit: MrBIOS]
[01:52:22] -!- guys has quit [Ping timeout: 246 seconds]
[01:53:32] -!- eschwartz has joined #archlinux32
[03:33:52] -!- eschwartz has quit [Ping timeout: 246 seconds]
[03:48:22] -!- eschwartz has joined #archlinux32
[03:58:48] -!- john__ has joined #archlinux32
[03:58:52] -!- DCyrax has quit [Read error: Connection reset by peer]
[04:10:37] -!- MrBIOS has joined #archlinux32
[04:15:28] -!- MrBIOS has quit [Ping timeout: 245 seconds]
[04:15:48] -!- MrBIOS has joined #archlinux32
[04:56:07] -!- MrBIOS has quit [Read error: Connection reset by peer]
[05:02:34] -!- MrBIOS has joined #archlinux32
[05:25:12] -!- MrBIOS has quit [Quit: MrBIOS]
[05:34:27] -!- eschwartz has quit [Ping timeout: 268 seconds]
[05:34:57] -!- MrBIOS has joined #archlinux32
[05:39:10] -!- MrBIOS has quit [Ping timeout: 246 seconds]
[05:48:15] -!- eschwartz has joined #archlinux32
[06:08:43] -!- slacka123 has quit [Remote host closed the connection]
[06:18:00] -!- MrBIOS has joined #archlinux32
[06:22:34] -!- MrBIOS has quit [Ping timeout: 252 seconds]
[07:34:24] -!- eschwartz has quit [Ping timeout: 258 seconds]
[07:47:44] -!- eschwartz has joined #archlinux32
[08:16:31] -!- dopsi has quit [Quit: ZNC - https://znc.in]
[08:17:28] -!- dopsi has joined #archlinux32
[08:18:49] -!- MrBIOS has joined #archlinux32
[08:23:28] -!- MrBIOS has quit [Ping timeout: 258 seconds]
[09:16:36] -!- deep42thought has joined #archlinux32
[09:16:36] <buildmaster> Hi deep42thought!
[09:16:36] <buildmaster> !rq deep42thought
[09:16:37] <phrik> buildmaster: <deep42thought> it was too slow - they don't offer that speed anymore, so they rounded down to 0
[09:36:39] -!- eschwartz has quit [Ping timeout: 244 seconds]
[09:43:54] -!- MrBIOS has joined #archlinux32
[09:47:19] -!- alexandre9099 has quit [Quit: ZNC 1.7.3 - https://znc.in]
[09:48:15] -!- MrBIOS has quit [Ping timeout: 258 seconds]
[09:49:17] -!- alexandre9099 has joined #archlinux32
[09:50:10] -!- eschwartz has joined #archlinux32
[10:15:53] -!- thePiGrepper has quit [Ping timeout: 245 seconds]
[10:16:57] -!- thePiGrepper has joined #archlinux32
[11:37:07] -!- eschwartz has quit [Ping timeout: 258 seconds]
[11:44:43] -!- MrBIOS has joined #archlinux32
[11:50:16] -!- eschwartz has joined #archlinux32
[11:57:23] <buildmaster> pentium4/startdde is broken (says nlopc46).
[11:57:43] <buildmaster> pentium4/deepin-dock is broken (says rechenknecht).
[12:01:00] <buildmaster> i686/deepin-dock is broken (says rechenknecht).
[12:01:14] <buildmaster> pentium4/virtualbox-modules-arch is broken (says nlopc46).
[12:12:58] -!- thePiGrepper has quit [Ping timeout: 245 seconds]
[12:35:03] <buildmaster> pentium4/sd is broken (says nlopc46).
[13:28:07] <buildmaster> pentium4/doublecmd is broken (says eurobuild3).
[13:28:32] <buildmaster> pentium4/fpc is broken (says nlopc46).
[13:36:04] <buildmaster> i686/sd is broken (says eurobuild3).
[13:36:54] -!- eschwartz has quit [Ping timeout: 244 seconds]
[13:38:55] <buildmaster> pentium4/librsvg is broken (says nlopc46).
[13:50:58] -!- eschwartz has joined #archlinux32
[13:57:54] <buildmaster> i686/startdde is broken (says rechenknecht).
[13:58:49] <buildmaster> i686/librsvg is broken (says nlopc46).
[14:01:06] <deep42thought> just FYI: the archive is archiving again, I set it up freshly on the master mirror - It still needs to finish copying over the older packages from tyzoid, though
[14:01:56] <buildmaster> i686/virtualbox-modules-arch is broken (says eurobuild3).
[14:09:09] -!- ofara_ has quit [Quit: ofara_]
[14:13:54] -!- thePiGrepper has joined #archlinux32
[14:29:01] -!- ofara_ has joined #archlinux32
[14:37:43] <buildmaster> pentium4/virtualbox is broken (says buildknecht).
[14:45:42] -!- ofara_ has quit [Quit: ofara_]
[14:52:08] -!- ofara_ has joined #archlinux32
[15:17:43] -!- eschwartz has quit [Ping timeout: 246 seconds]
[15:49:15] -!- MrBIOS has quit [Quit: MrBIOS]
[15:58:30] -!- abaumann has joined #archlinux32
[15:58:30] <buildmaster> Hi abaumann!
[15:58:30] <buildmaster> !rq abaumann
[15:58:31] <phrik> buildmaster: <abaumann> Maybe we should convert Rust to C using a Go to Vala transpiler written in Ocaml?
[15:58:40] <deep42thought> Hi abaumann!
[15:58:42] <abaumann> deep42thought: ah, thanks for re-archiving. :-)
[15:58:45] <abaumann> hi deep42thought
[15:58:56] <deep42thought> I'm searching through my package caches, currently
[15:59:18] <abaumann> for possible archive candidates?
[15:59:20] <deep42thought> yes
[15:59:34] <deep42thought> I have not emptied them in a long time, it seems
[15:59:36] <deep42thought> :-)
[15:59:42] <abaumann> in this case: good :-)
[16:00:02] <deep42thought> if you have any packages, that are missing, I can add those, too
[16:00:19] <abaumann> how to find them.. mmmh.
[16:00:54] <deep42thought> and how to send them to me :-D
[16:01:05] <deep42thought> this probably has both the same answer
[16:02:01] <abaumann> file lists..
[16:02:12] <abaumann> sort | uniq -c etc..
[16:02:23] <deep42thought> yeah
[16:02:40] <deep42thought> you can send me a file list (no hurry) and I'll send you what packages are missing
[16:03:21] <abaumann> ok.
[16:06:05] <deep42thought> ah, one thing though: the files from the cache do not come with a signature file (but that's not a big deal, I guess)
[16:06:45] -!- eschwartz has joined #archlinux32
[16:07:19] <abaumann> no.
[16:09:42] -!- quequotion has joined #archlinux32
[16:14:58] -!- thePiGrepper has quit [Ping timeout: 258 seconds]
[16:18:24] -!- deep42thought has quit [Quit: Leaving.]
[16:21:00] -!- abaumann has quit [Quit: leaving]
[16:26:24] -!- MrBIOS has joined #archlinux32
[16:55:33] -!- slacka123 has joined #archlinux32
[17:42:39] -!- MrBIOS has quit [Quit: MrBIOS]
[17:55:53] -!- isacdaavid has joined #archlinux32
[18:04:50] <isacdaavid> hey guys, i discovered while upgrading my i686 system that webkit2gtk (and all packages that depend on it by extension) has a broken dependency on icu 63
[18:05:05] -!- deep42thought has joined #archlinux32
[18:05:05] <buildmaster> Hi deep42thought!
[18:05:05] <buildmaster> !rq deep42thought
[18:05:06] <phrik> buildmaster: <deep42thought> golang-asvnp-acwduac-sefc-evfsx-vb4fs-sefvcy is broken (says me)
[18:05:17] <deep42thought> isacdaavid: yes, webkitgtk2 is not sse2-free
[18:05:21] <deep42thought> and thus cannot be built for i686
[18:05:34] <deep42thought> *webkit2gtk
[18:05:56] <deep42thought> it will be dropped from the i686 repositories once I have the courage to blacklist it
[18:09:20] <isacdaavid> shouldn't all the i686 repo be frozen at this point then? packages will continue breaking for existing installations as some dependencies fail to be updated
[18:10:38] <deep42thought> yes, some force update overwrote that logic ...
[18:11:07] <isacdaavid> i see
[18:11:27] <deep42thought> once the new pacman hits stable, most users will be upgraded to pentium4 anyways
[18:11:36] <deep42thought> then everything should be back to normal
[18:12:46] <isacdaavid> i welcome the transition
[18:16:09] <isacdaavid> and so will do #parabola, i think. all of the i686 hardware we care about (libreboot) is post-pentium 4
[18:16:57] <isacdaavid> btw, inkscape is broken too (bc of poppler). is that related to see2 too?
[18:17:16] <deep42thought> I have no clue
[18:17:52] <deep42thought> hmm, looks like pacman-5.1.3-1.4 does not automatically upgrade to pentium4 :-/
[18:18:33] <isacdaavid> :/
[18:19:25] -!- abaumann has joined #archlinux32
[18:19:26] <buildmaster> Hi abaumann!
[18:19:26] <buildmaster> !rq abaumann
[18:19:26] <phrik> buildmaster: <abaumann> I'm all for an docdepends upstreams and a doc hook, in order to reduce all those silly dependencies just to get documentation nobody reads afterwards, because they don't know about the 'man' command.. (rant of the day)
[18:20:12] <deep42thought> though the patch was applied :-/
[18:20:14] <deep42thought> hrrrm
[18:20:25] <abaumann> deep42thought: I collected 15980 packages (49B) from 12 machines or so. :-)
[18:20:38] <deep42thought> 49B?
[18:20:44] <abaumann> GB
[18:20:47] <abaumann> :-)
[18:20:48] <deep42thought> ah :-)
[18:21:04] <deep42thought> can you check out i686/testing/pacman?
[18:21:08] <deep42thought> that should have the pentium4 patch
[18:21:13] <abaumann> ok.
[18:21:16] <deep42thought> but happily installs i686 packages from my cache :-/
[18:21:51] <abaumann> mmh.
[18:22:48] <abaumann> isn't that what should happen?
[18:23:02] <deep42thought> no
[18:23:03] <abaumann> you have sse2 and set architecture=auto?
[18:23:07] <deep42thought> it should default to "pentium4"
[18:23:09] <deep42thought> yes
[18:23:30] <abaumann> ah. I'm testing on a non-sse2 machine, cool :-)
[18:23:37] <deep42thought> :-D
[18:23:44] <deep42thought> afk: dinner time
[18:23:55] <abaumann> ok. cu
[18:23:57] <abaumann> :-)
[18:25:14] -!- abaumann has quit [Quit: leaving]
[18:57:38] -!- AndrevS has joined #archlinux32
[19:03:04] -!- thePiGrepper has joined #archlinux32
[19:07:46] -!- thePiGrepper has quit [Ping timeout: 255 seconds]
[19:12:59] -!- MrBIOS has joined #archlinux32
[19:26:31] <deep42thought> on the vm, pacman does the right thing and refuses to install i686 packages
[19:34:37] <deep42thought> something about the sse2 detection is wrong ...
[19:34:42] <deep42thought> flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush acpi mmx fxsr sse sse2 ss tm syscall nx lm constant_tsc arch_perfmon rep_good cpuid pni monitor vmx est tm2 ssse3 cx16 xtpr sse4_1 popcnt rng rng_en ace ace_en ace2 phe phe_en pmm pmm_en lahf_lm tpr_shadow vnmi ida
[19:34:44] <deep42thought> there _is_ sse2
[19:34:51] <deep42thought> and uname -m reports "i686"
[19:34:58] <deep42thought> but pacman does not change it to pentium4 :-/
[19:35:27] <deep42thought> # pacman-conf Architecture
[19:35:27] <deep42thought> i686
[19:58:54] <deep42thought> should CHOST be i686-pc-linux-gnu or pentium4-pc-linux-gnu?
[20:32:16] <deep42thought> ah, right, it should not (found the same question for devtools32) :-)
[21:22:47] -!- abaumann has joined #archlinux32
[21:22:48] <buildmaster> Hi abaumann!
[21:22:48] <buildmaster> !rq abaumann
[21:22:48] <phrik> buildmaster: <abaumann> not to have http://archlinux32.microsoft.com suddendly in the list of mirrors. ;-)
[21:25:55] <deep42thought> __builtin_cpu_supports("sse2") gives 0 on my box, although sse2 appears in the /proc/cpuinfo flags
[21:26:15] <deep42thought> this seems the wrong way to test for sse2 availability
[21:26:25] <deep42thought> also: https://stackoverflow.com
[21:26:26] <phrik> Title: c - does gcc's __builtin_cpu_supports check for OS support? - Stack Overflow (at stackoverflow.com)
[21:26:53] <deep42thought> but this is mostly for the reverse issue: feature detected (because supported by hardware) but unavailable (because disabled by os)
[21:31:07] <abaumann> hi deep42thought
[21:31:11] <deep42thought> hi abaumann
[21:31:26] * deep42thought looks for his manners again
[21:31:39] <abaumann> :-)
[21:31:51] <abaumann> just sent you a list of archive packages (possible candidates)
[21:31:56] <deep42thought> ah, fine
[21:33:35] <abaumann> mmh. about __builtin_cpu_supports..
[21:33:46] <abaumann> ..is this a gcc intrinsic, so it detects SSE2 at compile time?
[21:34:01] <deep42thought> it is supposed to detect at runtime
[21:34:09] <abaumann> *abaumann has a look at the implementation..
[21:34:16] <deep42thought> but it does not matter - I'm compiling a small test program and get the same (wrong) result
[21:41:09] <abaumann> "In particular, the file containing the CPU detection code should be compiled without these options. "
[21:41:39] <deep42thought> ok, we're doing something wrong
[21:41:56] <abaumann> I wonder, what flags are set when pacman.c is compiled..
[21:41:56] <deep42thought> but I don't understand, why the same pacman works on the vm and works not on my box here :-/
[21:42:17] <abaumann> it might not do the same as /proc/cpuinfo
[21:42:36] <abaumann> but, I would suspect it detects SSE2 always on, not off
[21:42:40] <abaumann> would be more logical
[21:42:57] <deep42thought> yeah
[21:43:33] <abaumann> mmh. works fine in my test.
[21:44:30] <abaumann> http://archlinux32.andreasbaumann.cc
[21:44:37] <abaumann> compiled with 'gcc -o test test.c'
[21:44:58] <deep42thought> that's pretty much what I did, too
[21:45:30] <deep42thought> https://eckner.net
[21:45:57] <abaumann> mmh. strange. maybe some compiler flags?
[21:46:01] <deep42thought> $ gcc test.c && ./a.out
[21:46:03] <deep42thought> No SSE2
[21:46:12] <deep42thought> your source
[21:46:16] <abaumann> mmh.
[21:46:43] <abaumann> well. we could run some SSE2 inline assembly and trap on SIGILL ;-)
[21:46:46] <deep42thought> can you throw together a quick "fail on no-sse2" program?
[21:47:24] <abaumann> not sure.. something linking to qt/webkit? ;-)
[21:47:29] <deep42thought> LOL
[21:47:33] <abaumann> I'll try :-)
[21:47:36] <deep42thought> I had hoped for an easier way
[21:48:21] <abaumann> yeah. Something is lurking in my mind. I saw something like that lately..
[21:49:07] <abaumann> error: impossible constraint in ‘asm’asm("movdqa %xmm0, %0;" ^~~
[21:49:09] <abaumann> mmh.
[21:49:29] <abaumann> I'm not the biggest fan of gcc inline assembly syntax :->
[21:50:37] <abaumann> there was a has_sse2 test somewhere (the one craoking in chromium)
[21:52:00] <abaumann> I even wrote a patch once.. *grmpf* cannot find it
[21:52:46] <abaumann> mmh. somewhere in chromium, but checking out chromium is a pain..
[21:53:43] <abaumann> gallium..
[21:53:52] -!- thePiGrepper has joined #archlinux32
[21:54:31] <abaumann> so, this means checking out mesa.. much better :-)
[21:54:53] <abaumann> aha.. https://dri.freedesktop.org
[21:54:54] <phrik> Title: Gallium3D: gallium/auxiliary/util/u_cpu_detect.c Source File (at dri.freedesktop.org)
[21:55:07] <abaumann> ui. :-)
[21:55:21] <abaumann> yep, trapping SIGILL
[21:56:44] <abaumann> ah. that's for SSE
[21:56:53] <abaumann> sse2 goes over a sysctl.
[21:57:08] <abaumann> or cpuid.
[21:57:09] <abaumann> mmh.
[21:57:31] <deep42thought> should I just include this file and try?
[21:57:59] <abaumann> you can. but it does a little bit too much for our purpose.
[21:59:41] <abaumann> but actually: fopen( "/proc/cpuinfo" ) and parsing for "sse2" is also an option..
[22:00:45] <deep42thought> doesn't detect sse2, either
[22:00:50] <abaumann> ah, BSDs hat sysctl and mibs, now I recall.
[22:00:54] <abaumann> Not a Linux thing..
[22:02:27] <deep42thought> so maybe, it's deactivated in software, but still present in /proc/cpuinfo?
[22:02:37] <deep42thought> flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush acpi mmx fxsr sse sse2 ss tm syscall nx lm constant_tsc arch_perfmon rep_good cpuid pni monitor vmx est tm2 ssse3 cx16 xtpr sse4_1 popcnt rng rng_en ace ace_en ace2 phe phe_en pmm pmm_en lahf_lm tpr_shadow vnmi ida
[22:02:42] <deep42thought> ^just for reference
[22:02:59] <abaumann> I have almost a test program ready using cpuid
[22:03:22] <deep42thought> I'll happily test
[22:03:43] <deep42thought> it's ~1h until the clone army finally attacs, so: no hurry
[22:04:33] <abaumann> http://archlinux32.andreasbaumann.cc
[22:04:58] <abaumann> maybe you disable CPUID in the BIOS? ;-)
[22:04:59] <deep42thought> $ gcc test2.c && ./a.out
[22:04:59] <deep42thought> Has SSE2
[22:05:04] <abaumann> aha.
[22:05:15] <abaumann> a, better than gcc intrinsics..
[22:05:39] <abaumann> *abaumann is testing on a non-SSE2 machine, if it actually says 'No SSE2' :-)
[22:06:10] <deep42thought> $ gcc test2.c && ./a.out
[22:06:10] <deep42thought> No SSE2
[22:06:12] <deep42thought> yup
[22:06:16] <deep42thought> works on my alix, too
[22:06:56] <abaumann> it will not work, if SSE2 is disabled. so we could cascade the tests.
[22:07:24] <deep42thought> what do you mean?
[22:07:30] <deep42thought> "will not work" and "cascade"?
[22:07:48] <abaumann> there is a failure code, when cpuid is disabled or not available.
[22:07:57] <abaumann> then fallback to the builtin intrinsics
[22:08:01] <deep42thought> ah
[22:08:02] <deep42thought> ok
[22:08:06] <deep42thought> sounds reasonable
[22:08:18] <abaumann> and will break in corner cases for sure :-)
[22:08:35] <deep42thought> btw: I made pacman's python test suite parse /proc/cpuinfo to get the available flags
[22:09:11] <abaumann> idea: we make it interactive in the first call of pacman: "Pacman detected you system has SSE2, do you want to upgrade to get a blazing fast 32-bit machine? yes/no/maybe" ;-)
[22:09:39] <deep42thought> lol
[22:09:54] <deep42thought> !grab abaumann
[22:09:55] <phrik> deep42thought: Tada!
[22:11:13] <abaumann> AMD-K5: no SSE2, Pentium III, no SSE2, Alix AMD Geode, no SSE2, eeepc says has SSE2 .. looking good. :-)
[22:11:23] <deep42thought> :-)
[22:17:50] -!- quequotion has quit [Remote host closed the connection]
[22:17:54] <deep42thought> btw, abaumann: I sent you the list of packages missing in the archive
[22:18:00] <deep42thought> (but once more: no hurry)
[22:18:02] <abaumann> ah. thanks.
[22:18:23] <abaumann> true. it's only to fill up the gaps, right.
[22:18:29] -!- quequotion has joined #archlinux32
[22:18:30] <deep42thought> yeah
[22:18:41] <deep42thought> plus: I'm not done copying the stuff from tyzoid :-D
[22:18:46] <abaumann> mmh. the __builtin intrinsics should also call cpuid to detect SSE2..
[22:18:58] <abaumann> can imagine. It's quite big.
[22:19:11] <deep42thought> I hope it fits on the hdd
[22:19:12] <abaumann> If you say copy - aeh. isn't that a hard-linked thingy?
[22:19:12] <deep42thought> :-/
[22:19:32] <deep42thought> for current stuff it's hardlinks, but for the old stuff it's copying
[22:19:36] <abaumann> ah.
[22:19:41] <deep42thought> and I mean copy across the atlantic ...
[22:20:12] <abaumann> yeah.. the packages have to be filtered in prism.. ;-)
[22:20:23] <deep42thought> !grab abaumann
[22:20:24] <phrik> deep42thought: Tada!
[22:23:46] <abaumann> a de-duplicater and hard-linker: https://unix.stackexchange.com
[22:23:47] <phrik> Title: hard link - Is there an easy way to replace duplicate files with hardlinks? - Unix & Linux Stack Exchange (at unix.stackexchange.com)
[22:23:56] <abaumann> never tested, but sounds like a fun tool to use..
[22:24:08] <deep42thought> I have written something similar in bash
[22:24:12] <abaumann> ah. :-)
[22:24:13] <abaumann> cool
[22:24:29] <deep42thought> which works for my raid with >~20TB
[22:24:39] <abaumann> mmh..
[22:24:40] <deep42thought> and generates metadata in the order of 100GB
[22:24:47] <abaumann> wow.
[22:24:59] <deep42thought> list of inode -> hash
[22:25:17] <deep42thought> which can then be used to merge identical files
[22:27:54] <deep42thought> but deduplicating is useless for packages: if the file name differs, then the metadata differs and thus the pkgfile differs - and if the filename is the same, then it's already deduplicated, because packages are stored by-name
[22:28:12] <abaumann> ah. true again.
[22:34:10] -!- MrBIOS has quit [Ping timeout: 246 seconds]
[22:35:11] -!- MrBIOS has joined #archlinux32
[22:37:08] <abaumann> https://gcc.gnu.org __get_cpuid, also an option
[22:37:29] <abaumann> my version test2.c is shamelessly copied from mesa..
[22:37:46] <abaumann> yeah, but they basically just wrapc cpuid..
[22:39:00] <abaumann> so. the nigtly clones are all over me.. me sleep. :-)
[22:39:03] <abaumann> cu
[22:39:06] -!- abaumann has quit [Quit: leaving]
[22:39:59] <deep42thought> good night
[22:40:00] <deep42thought> !
[23:10:10] -!- deep42thought has quit [Quit: Leaving.]