#archlinux-ports | Logs for 2018-01-13

Back
[01:26:20] -!- krumelmonster has quit [Quit: Leaving.]
[02:00:08] -!- hringriin_ has joined #archlinux-ports
[02:01:55] -!- hringriin has quit [Ping timeout: 260 seconds]
[02:01:55] hringriin_ is now known as hringriin
[02:32:32] -!- sunshavi has joined #archlinux-ports
[04:06:10] <buildmaster> haskell-unixutils is broken (says buildknecht3).
[04:52:01] -!- dimbulb has quit [Ping timeout: 248 seconds]
[05:24:23] -!- hringriin_ has joined #archlinux-ports
[05:28:02] -!- hringriin has quit [Ping timeout: 276 seconds]
[05:28:02] hringriin_ is now known as hringriin
[07:02:30] -!- bill-auger has quit [Remote host closed the connection]
[07:06:24] -!- bill-auger has joined #archlinux-ports
[08:00:14] -!- deep42thought has joined #archlinux-ports
[08:03:07] <deep42thought> pandoc is still pending to be built, but can't, because it depends on haskell-hslua-module-text, which is currently broken
[08:08:26] <buildmaster> pandoc is broken (says buildknecht3).
[08:10:04] <deep42thought> jupp, can't be built now, because haskell-hslua-module-text needs to be rebuilt against the current haskell-hslua first ...
[08:24:41] -!- p71 has quit [Remote host closed the connection]
[08:28:32] -!- p71 has joined #archlinux-ports
[08:43:47] -!- abaumann has joined #archlinux-ports
[08:44:13] <abaumann> pandoc, haskell-hslua: the problem is failing tests in pandoc due to non-32-bit builds of haskell-hslua.
[08:44:33] <abaumann> but haskell-hslua either go into endless loops during check() or they fail
[08:44:59] <abaumann> forcing haskell-hslua without checks makes haskell-hslua-module-text fail in its tests.
[08:56:00] <deep42thought> yeah
[08:56:09] <deep42thought> so I skipped those tests, too (for now)
[08:57:13] <deep42thought> ok, pandoc builds, but its check() fails, too
[08:57:14] <deep42thought> :-/
[09:09:29] <buildmaster> pandoc-citeproc is broken (says buildknecht3).
[09:13:55] <deep42thought> this is what you get, when you build with packages from a tier1 mirror instead of the master mirror ...
[09:20:01] -!- deep42thought has quit [Quit: Leaving.]
[09:46:46] -!- oaken-source has joined #archlinux-ports
[10:06:47] <buildmaster> pjproject-savoirfairelinux is broken (says rechenknecht).
[10:49:32] -!- krumelmonster has joined #archlinux-ports
[11:05:13] -!- abaumann has quit [Quit: leaving]
[11:11:11] -!- krumelmonster has quit [Ping timeout: 248 seconds]
[11:50:47] -!- abaumann has joined #archlinux-ports
[11:51:03] <abaumann> deep42thought: mmh.. I'm building against tier1?
[11:58:45] <abaumann> damn. yes. because of i486 experiments..
[12:03:07] -!- abaumann has quit [Quit: leaving]
[12:10:50] <buildmaster> libfm is broken (says rechenknecht).
[12:26:24] <buildmaster> pdfmod is broken (says buildknecht2).
[13:09:56] -!- krumelmonster has joined #archlinux-ports
[13:20:05] -!- because has quit [Ping timeout: 268 seconds]
[13:24:58] -!- alyptik has joined #archlinux-ports
[13:26:07] -!- deep42thought has joined #archlinux-ports
[13:27:21] <deep42thought> abaumann_: I was referring to the broken pandoc-citeproc
[13:27:31] <deep42thought> I don't know which tier you build against ;-)
[13:37:19] -!- krumelmonster has quit [Read error: No route to host]
[13:39:01] -!- krumelmonster has joined #archlinux-ports
[13:46:13] -!- krumelmonster has quit [Quit: Leaving.]
[14:57:59] -!- abaumann has joined #archlinux-ports
[14:58:14] <abaumann> deep42thought: but it explained some other weird effects I had. :-)
[14:59:34] <deep42thought> I'm glad I could help :-D
[15:02:10] <deep42thought> It's a pity that build slaves running on (or close to) a mirror can't use that one due to the latency
[15:24:24] -!- deep42thought has quit [Quit: Leaving.]
[15:30:51] -!- Imsu has joined #archlinux-ports
[15:31:08] -!- Imsu has quit [Client Quit]
[15:37:12] -!- shentey has joined #archlinux-ports
[15:51:07] -!- abaumann has quit [Quit: leaving]
[16:39:28] -!- deep42thought has joined #archlinux-ports
[16:48:52] -!- buildmaster has quit [Remote host closed the connection]
[16:50:17] -!- buildmaster has joined #archlinux-ports
[16:59:22] -!- titus_livius has joined #archlinux-ports
[18:16:02] -!- isacdaavid has joined #archlinux-ports
[18:23:16] -!- bill-auger has quit [Ping timeout: 256 seconds]
[18:44:17] -!- shentey has quit [Quit: Konversation terminated!]
[19:32:20] -!- wCPO has joined #archlinux-ports
[20:30:31] -!- TomTheDragon has joined #archlinux-ports
[20:31:05] <TomTheDragon> Would it be inordinately difficult to port to i586?
[20:32:03] <deep42thought> abaumann_: ^
[20:32:19] <deep42thought> He's trying to :-)
[20:33:23] <TomTheDragon> hmm, thought I was the only one interested in such a thing
[20:33:45] <deep42thought> I'm interested, too. And abaumann_ is. But besides us three I know of noone
[20:34:38] <deep42thought> https://bbs.archlinux32.org
[20:34:40] <phrik> Title: i486 motivation thread / Artwork, Screenshots & Setups / Arch Linux 32 Forums (at bbs.archlinux32.org)
[20:34:55] <TomTheDragon> Mind you I haven't been an Arch user for years... mainly used Debian, but they have cut support so I will have to think about other options for my older boxes eventually.
[20:35:10] <deep42thought> huh?
[20:35:17] <deep42thought> debian cut support for some architectures?
[20:37:13] <TomTheDragon> yup
[20:37:33] <deep42thought> do you have any references?
[20:37:42] <TomTheDragon> supposedly they accidentally dropped i486 support a while back, and i586 very recently
[20:39:05] <TomTheDragon> https://www.debian.org
[20:39:07] <phrik> Title: Chapter 5. Issues to be aware of for stretch (at www.debian.org)
[20:40:31] <TomTheDragon> abaumann replied to that post and said it worked with an i686 build... interesting
[20:41:49] <deep42thought> yeah, I think it's the "i686 except no NOPL" category, which is also supported by debian
[20:42:04] <deep42thought> but anyway, he's trying to get real i486 running :-)
[20:42:46] <TomTheDragon> yeah. looks like he has some preliminary stuff, but no published buildscripts or anything
[20:43:22] <deep42thought> yeah, it's not yet "self containing"
[20:43:58] <TomTheDragon> hmm, you actually have to cross-compile for i486?
[20:44:10] <deep42thought> at least for bootstrapping
[20:44:28] <deep42thought> but he sounded like crosscompiling might be necessary later, too
[20:57:00] <TomTheDragon> interesting, mainly because I thought you could just pass -march flags
[20:58:04] <deep42thought> I have really no experience with that (abaumann has), but I think, the problem is, that many build-procedures guess the cpu or take this information from strange places
[20:58:20] -!- abaumann has joined #archlinux-ports
[20:58:25] <deep42thought> there he is :-)
[20:58:28] <deep42thought> Hi!
[20:58:40] <abaumann> hi there.
[20:59:08] <abaumann> just found iPXE had MME code in it because of Tivoli, so happy booting on my Unisys 486 :->
[20:59:28] <abaumann> now pxelinux won't coorperate.
[20:59:55] <abaumann> the preliminary stuff can boot a machine, but not much more.
[21:00:02] <TomTheDragon> hi abaumann. have wrangled with pxelinux before.
[21:00:09] <abaumann> especially building more packages on the i486 system is currently impossible.
[21:00:17] <abaumann> TomTheDragon: ah, good :-)
[21:00:23] <TomTheDragon> abaumann: yeah, I noticed, it was pretty minimal. why is it impossible?
[21:00:41] <abaumann> currently my cross-compiled make doesn't work.
[21:00:47] <TomTheDragon> hmm... ouch
[21:00:53] <abaumann> and I have no clue why it breaks.
[21:01:21] <TomTheDragon> quick question, not sure if you know, why do you have to cross-compile?
[21:01:47] <TomTheDragon> I thought that was only for architectures, not -march/-mtune
[21:02:44] <abaumann> yes, you can just use march=486, CalimeroTeknik did something like that some time ago for i586.
[21:03:13] <TomTheDragon> there is an advantage to cross compiling?
[21:03:13] <abaumann> But you have to be very careful.
[21:03:27] <abaumann> configure/autoconf behaves well under cross-compilation.
[21:03:34] <abaumann> no funny poking and probing of the host.
[21:03:49] <abaumann> most packages just see --host, --target etc. and behave nicely
[21:04:30] <TomTheDragon> Ah.
[21:05:09] <abaumann> I made several experiments: compile from an i486 version of Ubuntu, using a chroot with hacked uname architecture reporting.
[21:05:19] <abaumann> I was never really convinced to get a really stable system.
[21:05:57] <abaumann> I also found the hard way that qemu and 486 emulation is not the real thing.
[21:06:02] <TomTheDragon> Ah
[21:06:02] <abaumann> there are small subtle differences.
[21:06:21] <TomTheDragon> I imagine paths could mess with things, without cross compile
[21:06:21] <abaumann> for instance iPXE worked in qemu after taking out some MMX-based code..
[21:06:28] <abaumann> but it did't run on real hardware.
[21:06:42] <abaumann> oh, yes :-)
[21:08:14] <TomTheDragon> However.. best bet would probably be get an i486 bootstrap (stage0) type environment.
[21:08:29] <abaumann> but the iso boots, no ramdisk, no modules, no systemd. pacman has issues. make not working, gcc/glibc/toolchain seems to be fine, networing is ok (but for ping with an architecture mismatch). that's more or less the current status.
[21:09:01] <abaumann> the kernel on the iso boots in qemu/486
[21:09:04] -!- yar has quit [Ping timeout: 265 seconds]
[21:09:47] <TomTheDragon> no make, so you can't rebuild gcc?
[21:10:00] <abaumann> I cannot even compile a 'which' :-)
[21:10:07] <TomTheDragon> ouch
[21:10:19] <abaumann> I could use the make from crosstool-ng, but it also looked a little bit buggy.
[21:11:06] <deep42thought> can you try to compile make on the system itself?
[21:11:21] <deep42thought> or is the makefile too large to execute everything manually?
[21:11:25] <abaumann> yeah. not really. becuase make needs make. :-)
[21:11:43] <deep42thought> you can execute the commands in the makefile manually
[21:11:44] <abaumann> I could make my own let's say shell-script to build make.
[21:11:49] <abaumann> and then I have a working make.
[21:11:52] <deep42thought> right
[21:11:57] <abaumann> but make should really not break under cross-compilation.
[21:12:08] <deep42thought> should
[21:12:10] <abaumann> this usually means something is broken, either in make or in the toolchain.
[21:12:12] <TomTheDragon> it throws an invalid opcode?
[21:12:20] <abaumann> so scratch-and-sniff is at order.
[21:12:31] <abaumann> no, it just executes only internal rules, no external ones.
[21:12:51] <TomTheDragon> hmm, now that is odd
[21:12:57] <abaumann> as if run without a 'Makefile' file.
[21:13:46] <abaumann> usually using make in debug mode on a simple test package should show differences and hints.
[21:13:48] <TomTheDragon> Maybe the GNU stuff is legitimately broken in that version?
[21:13:53] <abaumann> just, I had no time so far. :-)
[21:14:02] <abaumann> not very likely.
[21:14:32] <abaumann> besides: others use make too. so, something is weird in my setup..
[21:15:00] <abaumann> mmh. ne2k drivers for the unisys cwd-4002, also not easy to find..
[21:15:55] <TomTheDragon> ne2k-isa driver?
[21:16:00] <abaumann> yep.
[21:16:23] <abaumann> well.. it's embedded on the board. I would guess it should be PCI?
[21:16:39] <TomTheDragon> hmm, it could be
[21:16:53] <TomTheDragon> didn't think PCI was around then
[21:16:56] <abaumann> n2k is a spec, not a card actually.
[21:17:16] <abaumann> so you have to use chipset or card maybe.
[21:17:50] <TomTheDragon> You did do multi-stage compile of GCC, right?
[21:17:58] <abaumann> yep.
[21:18:17] -!- deep42thought has quit [Quit: Leaving.]
[21:18:30] <abaumann> the cross-compiler is fine, I compiled glibc, gcc and binutils with the crosstools-ng to i486.
[21:18:40] <abaumann> and there I can assemble/compile test programs.
[21:18:50] <abaumann> I couldn't run the tests though.
[21:19:58] <abaumann> The unisys is a weird beast, it's a modern i486.
[21:20:11] <abaumann> all on-board and then an ISA-riser, but there is the soundcard only.
[21:20:24] <TomTheDragon> ah, I see
[21:20:42] <abaumann> well-modern. relatively. :-)
[21:22:54] -!- oaken-source has quit [Ping timeout: 256 seconds]
[21:23:57] <TomTheDragon> true
[21:32:02] -!- yar has joined #archlinux-ports
[21:32:18] <abaumann> aha. the isa ones are gone. n2k_pci.c is orphaned. :-)
[21:32:46] <TomTheDragon> I see
[21:32:50] <TomTheDragon> yuck
[21:33:33] <abaumann> but ipxe can boot and I can enable verbose PCI messages, so I should be able to see what the thing is underneath.
[21:33:44] <abaumann> I would bet it's PCI.
[21:40:39] <abaumann> ok. It's ISA, iPXE picks ne2k_isa.c
[21:42:29] <abaumann> ah. this explains my pxelinux freezes, wrong drivers. :-)
[21:43:25] <abaumann> I would actually prefer to use a modern ipxe script and boot the kernel and ramdisk directly. this is more the Archlinux style.
[21:43:41] <abaumann> tftp/pxelinux is a little bit cumbersome.
[21:49:44] <abaumann> ok.. continuation tomorrow. :-)
[21:49:45] -!- abaumann has quit [Quit: leaving]