#archlinux32 | Logs for 2018-11-16

Back
[00:05:03] -!- thePiGrepper has quit [Ping timeout: 245 seconds]
[00:06:05] <buildmaster> i686/emacs-haskell-mode is broken (says nlopc46).
[01:04:26] -!- MrBIOS has quit [Quit: MrBIOS]
[01:13:00] -!- MrBIOS has joined #archlinux32
[01:44:40] -!- MrBIOS has quit [Quit: MrBIOS]
[01:54:38] -!- woshty has quit [Ping timeout: 245 seconds]
[01:54:45] -!- davor_ has joined #archlinux32
[01:54:54] -!- davor has quit [Ping timeout: 244 seconds]
[01:54:59] davor_ is now known as davor
[03:38:50] -!- thePiGrepper has joined #archlinux32
[03:51:09] -!- girls has quit [Ping timeout: 244 seconds]
[03:51:40] -!- tyzoid has quit [Ping timeout: 244 seconds]
[04:00:07] -!- tyzoid has joined #archlinux32
[04:00:08] <buildmaster> Hi tyzoid!
[04:00:08] <buildmaster> !rq tyzoid
[04:00:08] <phrik> buildmaster: <tyzoid> bugs are just unintended features
[04:02:33] <buildmaster> i686/prettier is broken (says nlopc46).
[05:33:43] -!- NoobAlice1 has quit [Quit: Leaving.]
[05:58:28] -!- thePiGrepper has quit [Ping timeout: 272 seconds]
[05:59:47] -!- thePiGrepper has joined #archlinux32
[06:13:21] -!- girls has joined #archlinux32
[06:13:22] <buildmaster> Hi girls!
[06:13:22] <buildmaster> !rq girls
[06:13:22] <phrik> buildmaster: <girls> I'm married, I don't cook
[06:40:37] -!- thePiGrepper has quit [Ping timeout: 244 seconds]
[06:41:10] -!- thePiGrepper has joined #archlinux32
[06:47:52] -!- thePiGrepper has quit [Ping timeout: 272 seconds]
[06:59:46] -!- thePiGrepper has joined #archlinux32
[07:00:41] -!- djmoch has quit [Ping timeout: 260 seconds]
[07:12:58] -!- Cthulu201 has quit [Ping timeout: 245 seconds]
[07:14:45] -!- Cthulu201 has joined #archlinux32
[07:53:13] -!- T`aZ has quit [Ping timeout: 268 seconds]
[08:21:26] -!- deep42thought has joined #archlinux32
[08:21:26] <buildmaster> Hi deep42thought!
[08:21:26] <buildmaster> !rq deep42thought
[08:21:27] <phrik> buildmaster: <deep42thought> one day, some genius will prove, that all programming languages are isomorphic and there is actually only _one_ compiler
[09:54:16] -!- woshty has joined #archlinux32
[09:55:00] <buildmaster> i686/deepin-daemon is broken (says buildknecht2).
[09:59:38] <buildmaster> i686/systemd is broken (says buildknecht2).
[10:10:43] -!- Cthulu201 has quit [Read error: Connection reset by peer]
[10:22:51] -!- Cthulu201 has joined #archlinux32
[10:23:25] * buildmaster goes insane.
[10:25:00] <deep42thought> oh, pending mysql queries from yesterday :-D
[10:27:53] -!- abaumann has joined #archlinux32
[10:27:54] <buildmaster> Hi abaumann!
[10:27:54] <buildmaster> !rq abaumann
[10:27:56] <phrik> buildmaster: <abaumann> So I prefer cheap bootstrapping methods over the correct ones :-)
[10:27:56] <abaumann> hi deep42thought.
[10:27:59] <deep42thought> Hi abaumann
[10:28:16] <abaumann> your sed-foo looks complicated :-)
[10:28:27] <deep42thought> :-/
[10:28:31] <abaumann> for the so-detection.
[10:28:45] <abaumann> I'll have a while to understand it.
[10:28:52] <deep42thought> can you say in a few sentences how these versioned symbols work?
[10:29:05] <abaumann> no. not really.
[10:29:20] <abaumann> I'm not even sure, they are not on per-library basis.
[10:29:26] <deep42thought> they seem to be non-unique over all libraries
[10:29:27] <deep42thought> yeah
[10:29:38] * buildmaster resumes sanity.
[10:29:40] <deep42thought> many qt5 packages "provide" some version of qt5
[10:30:15] <deep42thought> for now, I add $libname-$symbolname-$symbolversion-$symbolhexversion as a "install_target"
[10:30:55] <deep42thought> e.g. libc.so.6-GLIBC_2.9-0x0d696919
[10:30:56] <abaumann> But I think, this will work just fine, when you extract all 'Version References'
[10:31:54] <abaumann> *abaumann: disclaimer, I'm not a great sed-guru. :-)
[10:32:16] <abaumann> from a logical standpoint the so-branch is ok, I think.
[10:32:57] <abaumann> Qt_5.11 and Qt_5. mmh ok.
[10:33:26] <abaumann> GLIBC_2.4, GLIBC_2.0, GLIBC_2.1.3, so we have to take the highest one..
[10:33:36] <deep42thought> the sed part is not, what I need you to check - it's my logic about how these symbols and versions work
[10:33:38] <abaumann> ..which means we have to know how to compare those funny versions?
[10:33:48] <abaumann> :-)
[10:33:54] <deep42thought> oh?
[10:33:57] <deep42thought> hmm :-/
[10:34:25] <deep42thought> no
[10:34:38] <deep42thought> libc.so.6 defines all kind of versions of GLIBC_*
[10:35:31] <deep42thought> have a look at: objdump -x /usr/lib/libc.so.6 | sed '/Version definitions/,/^$/!d'
[10:36:01] <deep42thought> it exports GLIBC_2.2.5 upto GLIBC_2.28
[10:36:22] <abaumann> so, there are linker scripts per version of glibc.
[10:37:20] <abaumann> mmh. isn't all this version tagging of symbols just there, so that you CAN change the name of symbols without breaking the ABI?
[10:37:50] <deep42thought> possibly
[10:38:08] <deep42thought> but I could also be, so that you can keep the soname and still change the ABI
[10:38:08] <abaumann> So, basically, the functions like as versioned variants all together in the same binary.
[10:38:21] <abaumann> for glibc this means, programs can use older versions of a function.
[10:38:45] <deep42thought> you mean, there is some ordering information in that versions?
[10:38:59] <abaumann> well. at least logically.
[10:39:07] <abaumann> I don't think from the ld/ldd's perspective.
[10:39:29] <abaumann> there it's just matching symbols having the same name: GLIBC_2.4@memcpy or so...
[10:39:31] <deep42thought> but it would need to know that ordering in order to be able to use a different version, wouldn't it?
[10:39:57] <deep42thought> or is it compile-time, you're talking about?
[10:40:05] <abaumann> yeah. I think the only time the linker does some version comparisions is in the name of the soname and .so.x.x.x
[10:40:11] <deep42thought> because that's nothing, the buildmaster needs to worry about
[10:40:45] <deep42thought> ^read this in german grammar
[10:40:53] <abaumann> no, the buildmaster just has to know that the version is Qt_5.11 and things depend on it, so it should not bump things with Qt_5.12
[10:41:08] <deep42thought> yeah
[10:41:08] <deep42thought> ok
[10:41:16] <abaumann> worth a shot. :-)
[10:41:47] <deep42thought> btw: I'm currently scrubbing the mirror's packages to extract the version information from all previously built packages
[10:41:52] <abaumann> *abaumann makes a note to read more about how linkers actually work..
[10:42:09] <deep42thought> .. so we should get a pretty complete picture of required and provided symbols
[10:42:28] <abaumann> interesting. so we have an early alert system for broken symbols. neat. :-)
[10:43:06] <deep42thought> ... just as we have one for broken linking in general now
[10:43:16] <deep42thought> (which seems to be not 100% working, though)
[10:44:01] <abaumann> there will always be things, we cannot detect.
[10:44:11] <deep42thought> !grab abaumann
[10:44:12] <phrik> deep42thought: Tada!
[10:44:22] <deep42thought> WIMPs, for example
[10:44:29] <abaumann> yes :-)
[10:45:29] <abaumann> or wheter orbital out-solar-system objects are actually a space ship or just a giant garlick bread stick.. ;-)
[10:45:42] <deep42thought> lol
[10:46:14] <abaumann> though I have to admit, I don't know how radar waves would reflect from a garlick bread stick..
[10:47:11] <deep42thought> the problem there is to have a radar system strong/sensitive enough to probe out-solar-system objects
[10:47:32] <abaumann> the garlick seems quite irrelevant in that equation ;-)
[10:47:36] * deep42thought imagines detonating nuclear bombs to create the necessary radar wave
[10:47:44] <abaumann> lol
[10:47:59] <deep42thought> this *might* even work
[10:48:33] <deep42thought> ... there was once a proposal to pump a laser by detonating (old soviet) nuclear bombs near to an optical cavity
[10:48:47] <deep42thought> luckily, it got rejected
[10:51:18] <abaumann> btw: I just pushed libnghttp2 (for nodejs/npm to work), let's see what this breaks..
[10:52:06] <deep42thought> I probably also broke some stuff by enthusiastically running 'db-update --progressive'
[10:52:41] <abaumann> we'll, I see it positively. We are a bleeding edge distro, so if it breaks, we fix it :-)
[10:52:43] <deep42thought> nice, short list: https://packages.archlinux32.org
[10:52:45] <phrik> Title:More and less critical issues with the database (at packages.archlinux32.org)
[10:53:10] <deep42thought> optimally, that list should be empty, though
[10:53:17] -!- Cthulu201 has quit [Ping timeout: 268 seconds]
[10:53:39] <abaumann> very good list.
[10:54:22] -!- Cthulu201 has joined #archlinux32
[10:54:23] <abaumann> I directly see some bugs I hit yesterday: plasma/KDE, x265 in ffmpeg during upgrade. x265 in vlc..
[10:55:15] <deep42thought> wait until I put all those symbols into our database - you'll see a lot more bugs, then ;-)
[10:55:26] <abaumann> jippie. :-)
[11:26:30] <buildmaster> i686/grafana is broken (says buildknecht2).
[11:32:40] -!- woshty has quit [Remote host closed the connection]
[11:33:22] -!- thePiGrepper has quit [Ping timeout: 268 seconds]
[11:45:03] -!- Cthulu201 has quit [Read error: Connection reset by peer]
[11:49:23] -!- Cthulu201 has joined #archlinux32
[12:13:52] -!- MrBIOS has joined #archlinux32
[12:33:51] -!- abaumann has quit [Quit: leaving]
[12:43:06] <buildmaster> i686/nasm is broken (says eurobuild3).
[12:53:57] <buildmaster> i686/dialog is broken (says nlopc43).
[12:56:38] -!- thePiGrepper has joined #archlinux32
[12:58:43] <buildmaster> i686/dbeaver is broken (says nlopc43).
[13:02:42] -!- MrBIOS has quit [Quit: MrBIOS]
[13:03:02] -!- MrBIOS has joined #archlinux32
[13:05:25] <deep42thought> strange, I have lots of gpg errors on my archlinux box
[13:05:32] <deep42thought> in the pacman keyring, that is
[13:07:07] <deep42thought> I lost an ultimately trusted key O.o
[13:08:20] -!- MrBIOS has quit [Ping timeout: 268 seconds]
[13:17:58] -!- woshty has joined #archlinux32
[13:27:30] -!- woshty has quit [Read error: Connection reset by peer]
[13:55:48] -!- NoobAlice has joined #archlinux32
[13:57:05] * buildmaster refuses to blacklist 19482 packages.
[13:57:05] <buildmaster> archlinux32: 1f97197ce09de8bc83e2d142bdb15500b825d3ca -> 71912a82f6019dde5795859f46be70430a40ac1f
[13:57:05] <buildmaster> community: bdd572409919ef872b5ece94cf71c2f5280ecdaf -> 4a0c3ca4c58bd324f2fd6501f8b7523313998a39
[13:57:05] <buildmaster> packages: 665e2055d5458ea5ac98204ff25b77368dcf2505 -> 3665c3d8149494a47fe02c840479cc8829169efb
[14:07:33] <buildmaster> i686/grim is broken (says eurobuild3).
[14:14:01] <deep42thought> damn
[14:14:11] <deep42thought> symbols can be in an archive
[14:14:20] <deep42thought> e.g. the filename does _not_ need to match
[14:37:45] -!- davor has quit [Quit: Bye]
[15:17:13] -!- Cthulu201 has quit [Quit: Nowhere special. I always wanted to go there.]
[15:21:11] -!- davor has joined #archlinux32
[15:21:16] -!- Cthulu201 has joined #archlinux32
[15:23:49] -!- djmoch has joined #archlinux32
[15:50:33] -!- guys has joined #archlinux32
[16:07:06] -!- guys has quit [Ping timeout: 272 seconds]
[16:21:23] -!- guys has joined #archlinux32
[16:33:39] -!- deep42thought has quit [Quit: Leaving.]
[18:07:16] -!- guys has quit [Ping timeout: 244 seconds]
[18:21:05] -!- guys has joined #archlinux32
[18:24:41] -!- autofsckk has quit [Ping timeout: 268 seconds]
[18:26:12] -!- autofsckk has joined #archlinux32
[18:53:48] -!- NoobAlice has quit [Ping timeout: 245 seconds]
[19:09:49] <buildmaster> i686/grim is broken (says nlopc46).
[19:12:46] <buildmaster> i686/vegeta is broken (says nlopc43).
[19:17:00] -!- MrBIOS has joined #archlinux32
[19:32:20] -!- MrBIOS_ has joined #archlinux32
[19:32:58] -!- MrBIOS has quit [Ping timeout: 245 seconds]
[19:32:58] MrBIOS_ is now known as MrBIOS
[19:43:09] -!- woshty has joined #archlinux32
[20:06:43] -!- guys has quit [Ping timeout: 245 seconds]
[20:21:44] -!- guys has joined #archlinux32
[20:43:20] <elibrokeit> hmm, interesting, tracking ABI symbols for packages?
[20:44:02] <elibrokeit> <abaumann> mmh. isn't all this version tagging of symbols just there, so that you CAN change the name of symbols without breaking the ABI?
[20:44:29] <elibrokeit> Well, not really -- its primary purpose is so that you can avoid breaking ABI at all, regardless of "reasons"
[20:44:46] <elibrokeit> and thereby also flag which version of glibc you will need
[20:45:42] -!- girls has quit [Ping timeout: 250 seconds]
[20:45:56] <elibrokeit> if we're talking other libraries, curl, openssl, etc, it's to prevent symbol clashes when using multiple versions of the library in one process -- because you're a fool of a closed-source developer, and you link to openssl 0.9.8 and also to the system libcurl
[20:46:08] -!- tyzoid has quit [Ping timeout: 250 seconds]
[20:52:11] -!- tyzoid has joined #archlinux32
[20:52:12] <buildmaster> Hi tyzoid!
[20:52:12] <buildmaster> !rq tyzoid
[20:52:13] <phrik> buildmaster: <tyzoid> As much as I hate willy-nilly disabling sanity / stability checks, but counterintuitively, I think disabling haskell's checks might make arch32 *more* stable.
[22:08:49] -!- guys has quit [Ping timeout: 252 seconds]
[22:23:24] -!- guys has joined #archlinux32
[22:33:21] -!- MrBIOS has quit [Read error: Connection reset by peer]
[22:33:29] -!- MrBIOS_ has joined #archlinux32
[23:08:41] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.2018-11-16T22:57:48.r3C68h.stdin"?.