#archlinux32 | Logs for 2019-04-10

Back
[00:46:55] -!- ubuntuxp has joined #archlinux32
[00:49:29] -!- samantaz_ has quit [Quit: Bye]
[01:09:01] -!- MrBIOS has quit [Quit: MrBIOS]
[01:59:41] -!- oaken-source has quit [Ping timeout: 268 seconds]
[02:12:03] -!- quequotion_ has joined #archlinux32
[02:13:28] -!- quequotion has quit [Ping timeout: 250 seconds]
[02:13:29] quequotion_ is now known as quequotion
[02:14:47] -!- elibrokeit has quit [Ping timeout: 240 seconds]
[02:15:20] -!- elibrokeit has joined #archlinux32
[02:19:32] -!- yans has joined #archlinux32
[03:06:48] <slacka123> Are plans still on track for a pentium4? Both my systems support SSE2. Any help needed for testing?
[03:42:07] -!- ubuntuxp has quit [Quit: ERC (IRC client for Emacs 26.1)]
[03:56:46] -!- quequotion has quit [Quit: quequotion]
[04:01:12] -!- yans has quit [Quit: chaos is the only true answer]
[04:46:14] -!- yans has joined #archlinux32
[04:56:58] -!- Dimtree has quit []
[05:27:12] -!- Dimtree has joined #archlinux32
[06:08:52] -!- deep42thought has joined #archlinux32
[06:08:52] <buildmaster> Hi deep42thought!
[06:08:52] <buildmaster> !rq deep42thought
[06:08:53] <phrik> buildmaster: <deep42thought> documented => the bug becomes a feature
[06:09:22] <deep42thought> slacka123: testing pentium4 is simpler than /really/ testing i686
[06:09:40] <deep42thought> that's the main reason, why we haven't gone forward to implement it yet ...
[06:30:29] <deep42thought> so if you come up with an good idea or even an implementation how to reliably test, that a given package does not use illegal instructions (for i686), then we're happy to hear/use it :-)
[07:32:55] -!- yans has quit [Ping timeout: 246 seconds]
[07:58:45] <slacka123> deep42thought: why can't you trust the compiler? "-march=pentium4" should not produce illegal instructions
[08:00:04] <slacka123> Do you test i486 or i686 for illegal instructions now?
[08:07:46] <deep42thought> the emphasis is on "should"
[08:08:00] <deep42thought> we do test very sporadic
[08:08:17] <deep42thought> e.g. most users have one or the other feature of pentium4 available but still run i686
[08:15:54] <deep42thought> actually, we could start adding pentium4 right now ... maybe we should - and sort out all the problems later :-D
[08:23:12] -!- deep42thought has quit [Quit: Leaving.]
[08:29:08] -!- oaken-source has joined #archlinux32
[08:40:45] -!- yans has joined #archlinux32
[08:44:12] -!- oaken-source has quit [Ping timeout: 245 seconds]
[09:18:47] -!- deep42thought has joined #archlinux32
[09:18:47] <buildmaster> Hi deep42thought!
[09:18:47] <buildmaster> !rq deep42thought
[09:18:48] <phrik> buildmaster: <deep42thought> doesn't one usually try to avoid rust on metals?
[10:04:30] -!- AndrevS has joined #archlinux32
[10:43:55] <wyre> KitsuWhooa: We could not remove the modules, cause cannot find the modules https://imgur.com
[10:43:57] <phrik> Title: Imgur: The magic of the Internet (at imgur.com)
[10:50:10] <KitsuWhooa> wyre: ah :(
[10:50:58] <KitsuWhooa> try dmesg | grep frame
[10:51:01] <KitsuWhooa> and see if anything comes up there
[11:10:46] <wyre> ok, I'll tell you 😉
[11:31:42] -!- oaken-source has joined #archlinux32
[12:15:09] -!- oaken-source has quit [Ping timeout: 255 seconds]
[12:36:50] -!- oaken-source has joined #archlinux32
[13:09:58] -!- oaken-source has quit [Ping timeout: 246 seconds]
[13:22:18] -!- oaken-source has joined #archlinux32
[13:32:04] -!- AndrevS has quit [Remote host closed the connection]
[13:43:20] -!- oaken-source has quit [Quit: Lost terminal]
[13:54:43] -!- lodger has joined #archlinux32
[13:58:56] -!- lodger has quit [Client Quit]
[15:04:39] -!- oaken-source has joined #archlinux32
[15:08:54] <slacka123> deep42thought: come to think of it, my dad has a real pentium4. Assuming it boots off a flash drive, I'd be happy to test it out.
[15:11:36] <slacka123> if most user have pentium4 capable cpu and it fixes some broken packages, it should be enabled
[15:12:01] <slacka123> I will test on a real pentium 4
[15:12:21] <deep42thought> I'll look into this friday, I think
[15:12:36] <deep42thought> but I'd like to get a second opinion from abaumann before I do :-)
[15:12:59] <deep42thought> I think, not much can go wrong - the worst case will be, that we have i686 and pentium4 with identical packages
[15:13:28] -!- oaken-source has quit [Ping timeout: 250 seconds]
[15:13:39] <deep42thought> thus, some "legal" on pentium4 but "illegal" for i686 and some others legal on i686 and sub-optimal on pentium4
[15:14:02] <deep42thought> the interesting part then will be to set the corretct build flags en-masse ...
[15:14:15] <slacka123> yes, and most users would benefit from that
[15:14:24] <slacka123> I will test it out
[15:37:49] <bill-auger> slacka123: im curious, why would it need to boot from a USB flash memory?
[15:38:05] <deep42thought> bill-auger: to install arch?
[15:38:23] <bill-auger> is that the only possible boot medium
[15:38:34] <deep42thought> it's the most convenient one
[15:38:40] <bill-auger> i still make the parabola ISOs ISO9660 compliant
[15:38:46] <deep42thought> well, "convenient" depends on your taste, though
[15:39:13] <bill-auger> its not the most convenient if the target computer was not designed for it
[15:39:19] * deep42thought hasn't booted from a spinning optical medium for like a decade now
[15:39:30] <deep42thought> bill-auger: true
[15:39:43] <deep42thought> it's also not the most convenient if you want to boot 100 machines from it ;-)
[15:43:44] <bill-auger> im confused though, why is a P4 an interesting target for the 486 port - i would think that the interesting tests would be on a 486 or 586 machine
[15:44:21] <deep42thought> no, not the i486 port
[15:44:32] <deep42thought> the plan is to have i486, i686 and pentium4
[15:44:45] <bill-auger> ok i assumed that conversation was rolled overf from yesterday
[15:44:58] <deep42thought> I assumed so, too
[15:45:17] <deep42thought> this is the plan: https://archlinux32.org
[15:45:18] <phrik> Title: Arch Linux 32 - Architecture Overview (at archlinux32.org)
[15:46:04] <bill-auger> ok cause i have a few machines in that class - those machines do not boot from USB, nost could not boot from CD either
[15:46:20] <deep42thought> in which class?
[15:46:26] <bill-auger> in many cases you need to boot from a floppy with GRUB on it then chainload to the CD
[15:46:42] <bill-auger> old pentiums
[15:53:39] <deep42thought> I have to go now, cu later
[15:54:13] -!- deep42thought has quit [Quit: Leaving.]
[16:45:59] -!- abaumann has joined #archlinux32
[16:45:59] <buildmaster> Hi abaumann!
[16:45:59] <buildmaster> !rq abaumann
[16:46:00] <phrik> buildmaster: <abaumann> wait... go-go-gadgeto-i686-test-machine
[16:46:55] <abaumann> deep42thought: no problems when adding the pentium4 branch of packages. makepkg would have the proper march flags, I presume. Bootstrapping from i686 is no problem. The only issue I see is too few build slaves on the horizon.. :-)
[16:51:34] <abaumann> completly unrelated: my Dallas CMOS and my sockets arrived.. I think, I go for the "clipping the pins" method, as I don't want to preserve the chip, but very much the motherboard!
[16:51:37] -!- oaken-source has joined #archlinux32
[16:51:39] <abaumann> gonna be fun. :-)
[16:52:51] <abaumann> deep42thought: the i686 parts would be the if [ "${CARCH}" = "i486" -o "${CARCH}" = "i686" ]; then
[16:55:57] -!- abaumann has quit [Quit: leaving]
[17:13:23] -!- isacdaavid has joined #archlinux32
[17:26:44] <slacka123> bill-auger: it's my dad's PC. I installed Arch 32 to a USB flash drive, so I can just take my OS + data over anytime and use it there. But I can't change his OS on him.
[17:31:45] <bill-auger> ok slacka123 - i was just noting that a pentium 4 almost certainly can boot from a CD - so you could still test it out even without your USB stick
[17:50:16] -!- isacdaavid has joined #archlinux32
[18:07:55] -!- AndrevS has joined #archlinux32
[18:30:12] <buildmaster> Hi deep42thought!
[18:30:12] <buildmaster> !rq deep42thought
[18:30:13] <phrik> buildmaster: good night, python - cu tomorrow python2!
[18:30:37] <deep42thought> ah, right, we should first deploy new devtools32 with build-staging-pentium4 and alike ...
[18:30:55] <deep42thought> this should care for the right build flags to makepkg
[18:31:15] <deep42thought> umm, I meant "staging-pentium4-build"
[18:51:17] -!- MrBIOS has joined #archlinux32
[19:23:44] -!- ubuntuxp has joined #archlinux32
[19:34:00] -!- samantaz has joined #archlinux32
[19:38:41] -!- titus_livius has joined #archlinux32
[19:56:00] <deep42thought> pentium4-pc-linux-gnu is not a valid CHOST, is it?
[19:56:09] <deep42thought> also CARCH=pentium4 sounds plain wrong :-/
[19:56:34] <deep42thought> -march=pentium4 should work, right?
[19:56:46] * deep42thought is still a novice to compiler flags
[19:59:41] <deep42thought> abaumann: If you have some spare time (lol), can you please have a look at the pentium4 branch of devtools32 - do the flags look sane?
[20:48:29] -!- abaumann has joined #archlinux32
[20:48:30] <buildmaster> Hi abaumann!
[20:48:30] <buildmaster> !rq abaumann
[20:48:30] <phrik> buildmaster: <abaumann> very soon only a machine learning algorithm will be able to devise a build plan for a Linux distribution..
[20:48:35] <abaumann> hi deep42thought
[20:48:45] <deep42thought> Hi abaumann
[20:49:49] <abaumann> -march=pentium4 -mtune=generic sounds right
[20:50:17] <deep42thought> what about pentium4-pc-linux-gnu?
[20:50:21] <abaumann> What is pacman doing with 'Architecure=pentium4'? Is it patched?
[20:50:30] <abaumann> ah, where is 'pentium4-pc-linux-gnu' used?
[20:50:31] <deep42thought> it's literally only a string
[20:50:40] <deep42thought> it's set as CHOST
[20:50:43] <deep42thought> no idea who uses that
[20:50:47] <abaumann> ah. right.
[20:50:49] <deep42thought> (that's why I'm asking you)
[20:50:50] <abaumann> CHOST="i686-pc-linux-gnu"
[20:51:17] <deep42thought> but CARCH=pentium4 does not break autoconf and friends?
[20:52:08] <abaumann> another option is to keep it as i686-pc-linux-gnu, as most people assume that to be the SSE2 i686 anyway.
[20:52:36] <deep42thought> what does that quadruple actually mean? who refers to it for what?
[20:53:00] <abaumann> historically this is the output of guess.config (or guess.sub), a shell sniffing tools in the autotools.
[20:53:12] <abaumann> It's mainly there to identify platforms for cross-compiling
[20:53:22] <abaumann> So you can set host, target, etc.
[20:53:49] <abaumann> I doubt it is passed to configure..
[20:53:57] <abaumann> maybe it's used in some PKGBUILDs to do that?
[20:54:04] <abaumann> *abaumann kicks a big grep..
[20:54:10] <deep42thought> but the first part might end up behind some -march=, right?
[20:54:37] <deep42thought> or might be interpreted by some compiler just like it was passed as -march=...
[20:54:39] <deep42thought> ?
[20:54:42] <abaumann> yeah. I have in mind it's cpu-system/platform-operatingsystem-vendor.
[20:55:09] <deep42thought> the question is, what valid values for the first component are
[20:55:22] <deep42thought> if "pentium4" is valid by any chance, we should use that
[20:55:25] <deep42thought> and "i686" if not
[20:55:28] <abaumann> I doubt..
[20:55:51] <abaumann> ..and it's likely to trouble some scripts..
[20:57:07] <abaumann> basically, to be mean, this lavel was not invented to do some micro-optimization for Intel chipsets matrixes. :-)
[20:57:13] <abaumann> *label.
[20:57:25] <abaumann> or, it stopped at i686. :-)
[20:57:44] <deep42thought> but we should change CARCH, because otherwise we could not have $CARCH dependent switches in the PKGBUILD
[20:57:45] <abaumann> https://www.gnu.org
[20:57:47] <phrik> Title: GNU gettext utilities: config.guess (at www.gnu.org)
[20:57:49] <deep42thought> will this cause any troubles?
[20:58:20] <abaumann> packages/llvm/repos/extra-x86_64/PKGBUILD: -DLLVM_HOST_TRIPLE=$CHOST \
[20:58:28] <abaumann> mmh. yes. I would keep CHOST as i686..
[20:58:36] <abaumann> CARCH.. hum..
[20:58:47] <abaumann> *abaumann kicks another grep on a rather slow disk.
[20:59:09] <abaumann> yeah. This one we need, otherwise we cannot distinguish between i486, i686 and pentium4 in PKGBUILD ifs.
[20:59:28] <abaumann> $([[ $CARCH == x86_64 ]] && echo --with-sse2 or so..
[20:59:33] <abaumann> *pentium4
[20:59:35] <abaumann> I meant.
[20:59:41] <deep42thought> yeah
[20:59:45] <deep42thought> that would be nice
[21:00:03] <deep42thought> but the question is, if (m)any scipt(s) will trip on funny CARCHs
[21:01:10] <abaumann> I think, most PKGBUILDs checked for x86_64 and i686 only in the past (not sure about Alarm, if they use it for ARMv6, ARMV7 checks, but I thought, they have a separate fine-grained architecture variable for this, keep CARCH to something generic).
[21:01:32] <abaumann> packages/libaio/trunk/PKGBUILD: CFLAGS="-march=${CARCH/_/-} -mtune=generic -O2 -pipe"
[21:01:55] <abaumann> here is an example of having CARCH=pentium33 actually really being helpful.
[21:02:25] <abaumann> packages/pyalpm/repos/extra-x86_64/PKGBUILD: PYTHONPATH="$PWD/build/lib.linux-$CARCH-3.7" pytest
[21:02:43] <abaumann> This one might just be to separate build directories by architecture, so probably also fine.
[21:03:07] <abaumann> yeah. those are more or less the three cases I can spot for CARCH
[21:03:55] <abaumann> ./config.sub pentium4-pc-linux-gnu => i786-pc-linux-gnu
[21:03:57] <abaumann> :-)
[21:04:12] <abaumann> so, the autotools canonizer goes for i786-pc-linux-gnu.
[21:04:30] <abaumann> mmh. I never tries a cross-compilation with gcc for instance with --host=i786-pc-linux-gnu
[21:04:40] <abaumann> would be interersting to see what happens.. :-)
[21:05:09] <abaumann> ./config.sub pentium3-pc-linux-gnu -> i686-pc-linux-gnu
[21:05:29] <abaumann> ./config.sub pentium2-pc-linux-gnu -> i686-pc-linux-gnu
[21:05:39] <abaumann> ./config.sub pentium-pc-linux-gnu -> i586-pc-linux-gnu
[21:05:41] <abaumann> mmh. almost..
[21:06:27] <deep42thought> ok, so we go with:
[21:06:27] <deep42thought> CARCH="pentium4"
[21:06:27] <deep42thought> CHOST="i686-pc-linux-gnu"
[21:06:27] <deep42thought> CFLAGS="-march=pentium4 -mtune=generic -O2 -pipe -fno-plt"
[21:06:27] <deep42thought> CXXFLAGS="-march=pentium4 -mtune=generic -O2 -pipe -fno-plt"
[21:06:29] <deep42thought> ok?
[21:06:39] <abaumann> I would say so.
[21:06:46] <abaumann> If not, you can blame it on me. :-)
[21:07:15] <deep42thought> I will do anyways :-D
[21:07:21] <abaumann> good. :-)
[21:07:22] <deep42thought> but thanks for making it official
[21:07:26] <deep42thought> :-D
[21:08:54] <abaumann> We could take 'flags' from '/proc/cpuinfo' to produce something like i686_fpu_mmx_sse_sse2, just to make the build matrix bigger. ;-)
[21:09:37] <abaumann> Something like the 'CPU Cambrian Explosion'..
[21:09:51] <deep42thought> lol
[21:10:41] <deep42thought> I pushed a new version of the pentium4 branch
[21:10:48] <abaumann> Interesting discussion in Slackware about SSE2: https://www.linuxquestions.org's-roadmap-4175533741/
[21:10:49] <phrik> Title: sse2 dependency for qt5; Slackware's roadmap? (at www.linuxquestions.org)
[21:10:51] <deep42thought> if it looks ok to you, I'll ff master to it
[21:11:16] <abaumann> ok.
[21:11:34] <abaumann> note the date of the discussion..
[21:14:09] <deep42thought> 4 years ago
[21:28:55] <deep42thought> oh, I totally missed the release of the black-hole "photograph"
[21:29:49] <abaumann> oh. true.
[21:30:21] <deep42thought> although I have to admit, the one in "interstellar" was more impressive, because it was in higher resolution ;-P
[21:34:23] <deep42thought> new devtools32 are available in [releng], devtools on archlinux32 should be built any time soon, now
[21:34:32] <abaumann> oh goodie. :-)
[21:34:42] <deep42thought> then we can experiment a little with staging-pentium4-build and friends :-)
[21:35:21] <abaumann> yep. but for me.. tomorrow.. I'm quite tired.. and I have to watch some unrealistic animations in documentaries about the new black hole picture..
[21:35:34] <deep42thought> np
[21:35:36] <deep42thought> no hurry
[21:35:57] <deep42thought> I've had enough lectures about black holes :-)
[21:36:06] <deep42thought> I won't watch a documentary about one
[21:36:06] <abaumann> can imagine :-)
[21:36:27] <deep42thought> or maybe I will, to make some fun about the animations and descriptions :-D
[21:36:35] <abaumann> exactly. :-)
[21:38:38] <abaumann> ok. cu.
[21:38:40] -!- abaumann has quit [Quit: leaving]
[21:38:55] -!- deep42thought has parted #archlinux32
[22:40:00] -!- yans has quit [Ping timeout: 246 seconds]
[22:40:52] -!- thePiGrepper has quit [Ping timeout: 245 seconds]
[23:17:30] -!- AndrevS has quit [Remote host closed the connection]