#archlinux32 | Logs for 2018-09-01

Back
[01:11:45] -!- woshty has quit [Ping timeout: 244 seconds]
[01:23:29] -!- bulletmark has joined #archlinux32
[01:57:25] -!- MrBIOS has quit [Quit: MrBIOS]
[03:04:56] -!- dopsi has quit [Quit: ZNC - https://znc.in]
[03:05:11] -!- dopsi has joined #archlinux32
[03:23:00] -!- ofara_ has quit [Ping timeout: 252 seconds]
[03:25:04] -!- ofara_ has joined #archlinux32
[08:30:01] -!- ofara has quit [Ping timeout: 246 seconds]
[08:33:53] * buildmaster resumes sanity.
[08:34:19] <buildmaster> firefox-developer-edition is broken (says buildknecht).
[09:13:50] -!- oaken-source has joined #archlinux32
[09:44:14] -!- abaumann has joined #archlinux32
[09:44:14] <buildmaster> Hi abaumann!
[09:50:38] <abaumann> thanks for fixing the buildmaster..
[09:50:38] <buildmaster> np, abaumann
[09:52:49] <abaumann> testing packages is zero? deep42thought, are you there? ;-)
[09:54:13] -!- oaken-source has quit [Ping timeout: 245 seconds]
[10:01:10] <abaumann> lib32-llvm in the database? huh?
[10:04:25] <abaumann> I think, the logic in the buildmaster must be changed: too often it tries to build packges, which make no sense to build, because they failed and are not blocking dependencies, e.g. why should the buildmaster build now riscv stuff with a backlog of 1200 packages?
[10:13:37] -!- abaumann has quit [Quit: leaving]
[10:25:08] -!- abaumann has joined #archlinux32
[10:25:09] <buildmaster> Hi abaumann!
[10:26:58] -!- abaumann has quit [Client Quit]
[10:47:16] * buildmaster goes insane.
[10:53:02] -!- deep42thought has joined #archlinux32
[10:53:02] <buildmaster> Hi deep42thought!
[10:53:16] <deep42thought> abaumann: yes, I fixed the buildmaster and marked all packages in testing as tested
[10:53:21] <deep42thought> so they should be moved if possible
[10:53:26] <deep42thought> (less forceful than -f)
[11:51:02] <deep42thought> abaumann: I'm open for suggestions in what direction to change the logic of get-assignment
[11:53:05] <deep42thought> currently it's: 1. requested, 2. toolchain (in respective order), 3. priority, 4. last_trial, 5. architecture matches _exactly_ the one of the slave, 6. is not part of loop, 7. source commit time, build_assignment.id
[11:54:07] <deep42thought> and only packages whose dependencies are all met or which are part of a loop are considered at all
[11:57:03] -!- abaumann has joined #archlinux32
[11:57:03] <buildmaster> Hi abaumann!
[11:57:06] <abaumann> hi deep42thought
[11:57:07] <deep42thought> Hi abaumann!
[11:57:10] <abaumann> welcome back :-)
[11:57:16] <deep42thought> thanks :-)
[11:57:23] <deep42thought> The Alps are nice.
[11:57:29] <abaumann> :-)
[11:57:56] <abaumann> yeah. but now the weather has changed.
[11:58:00] <abaumann> not so nice currently.
[11:58:13] <deep42thought> yes, was the same in Austria the last few days
[11:58:49] <deep42thought> but I don't really hate rain, so it was ok
[11:58:56] <abaumann> :-)
[12:00:36] <deep42thought> you removed your temp. lists on the build master again?
[12:00:46] <abaumann> which ones?
[12:00:51] <abaumann> the python lists?
[12:01:00] <deep42thought> bin/x bin/python-list bin/better-python-list
[12:01:12] <abaumann> yes, they got overwritten with a '0' in them.
[12:01:22] <abaumann> so they are not of much use anymore :-)
[12:01:33] <deep42thought> yeah, maybe a design flaw?
[12:01:46] <abaumann> I didn't think about it.
[12:01:53] <deep42thought> all the scripts that take a file of something as argument overwrite it by the result
[12:02:00] <abaumann> and I trigger far too many packages anyway because of a grep error *grmpf*
[12:02:11] <deep42thought> this is handy for interpret-email, because then it can print in the log how many pakcages were affected
[12:02:23] <abaumann> ah. that's why.
[12:02:37] <abaumann> the priority list looks ok now.
[12:02:43] <deep42thought> I usually use <(echo 'bla')
[12:02:59] <deep42thought> then it prints to stdout
[12:03:00] <abaumann> The following packages appear on the build- and deletion-list:
[12:03:00] <abaumann> lib32-llvm-libs
[12:03:00] <abaumann> lib32-llvm-libs
[12:03:00] <abaumann> lib32-llvm
[12:03:02] <abaumann> lib32-llvm
[12:03:05] <abaumann> SANITY CHECK FAILED
[12:03:07] <abaumann> mmh.
[12:03:10] <deep42thought> I removed them from the build list
[12:03:12] <abaumann> that was me with triggering 'llvm'
[12:03:16] <deep42thought> should be fixed in a few minutes
[12:03:21] <deep42thought> oh
[12:03:25] <deep42thought> yeah, right
[12:03:25] <abaumann> sometimes one would like to use NOT a regex :-)
[12:03:36] <deep42thought> seed-build-list does not apply the black listing afterwards
[12:04:12] -!- z3ntu has quit [Remote host closed the connection]
[12:04:31] <deep42thought> lunch, prepare for lunch!
[12:04:58] <abaumann> just finished with another i486 rebuild. Some 15 packages are failing, all because of glibc 2.28/gnulib and other ABI changes.
[12:05:10] <abaumann> oh.
[12:05:11] <abaumann> right.
[12:05:15] <abaumann> my curry is boiling..
[12:05:27] <abaumann> and my rice..
[12:05:30] <abaumann> ok. :-)
[12:05:31] <abaumann> cu later.
[12:12:19] -!- z3ntu has joined #archlinux32
[12:18:00] * buildmaster resumes sanity.
[12:38:48] -!- oaken-source has joined #archlinux32
[12:53:25] <deep42thought> abaumann: did you see this: https://bbs.archlinux32.org ?
[12:53:27] <phrik> Title:community/chromium / Creating/Maintaining Packages / Arch Linux 32 Forums (at bbs.archlinux32.org)
[13:00:39] <abaumann> mmh. now. thanks.
[13:00:46] <abaumann> *no
[13:01:06] <deep42thought> btw: the crazy handing out of assignments seems to be a bug: https://packages.archlinux32.org
[13:01:08] <phrik> Title:Arch Linux 32 - List of Next Package Builds (at packages.archlinux32.org)
[13:01:14] <deep42thought> this only shows 2 packages - most probably wrong
[13:01:36] <abaumann> they got stuck on some slaves?
[13:01:43] <deep42thought> no
[13:01:50] <deep42thought> those are ever-broken gcc-riscv-64 packages
[13:01:58] <abaumann> ah.
[13:02:00] <deep42thought> but the other packages should not be blocked by them
[13:02:09] <abaumann> exactly
[13:02:14] <deep42thought> e.g. the problem is, that the buildmaster sees missing dependencies for e.g. perl
[13:02:52] <deep42thought> buildmaster: why dont you build perl
[13:02:59] <buildmaster> deep42thought: "perl" has unmet dependencies:
[13:02:59] <buildmaster> perl
[13:02:59] <buildmaster> util-linux
[13:03:08] <abaumann> funny. :-)
[13:03:27] <deep42thought> buildmaster: why don't you build util-linux
[13:03:30] <buildmaster> deep42thought: "util-linux" has unmet dependencies:
[13:03:30] <buildmaster> perl
[13:03:30] <buildmaster> util-linux
[13:03:41] <deep42thought> so it should be a loop ...
[13:03:42] <deep42thought> hmmm
[13:03:50] <abaumann> IMHO most dependencies to scripting languages like perl, python should be an optdepend.
[13:03:56] <abaumann> especially for something like util-linux
[13:04:27] <deep42thought> is it required for compilation?
[13:04:32] <deep42thought> then it _must_ be a makedepends
[13:04:42] <deep42thought> but maybe we should not wait for (some) makedepends
[13:05:00] <deep42thought> e.g.: does it really matter that you used the latest git to check out your sources?
[13:05:35] <abaumann> util-linux has python, not perl as dependency (a makedepend)
[13:05:55] <abaumann> perl needs no util-linux.
[13:06:26] <deep42thought> util-linux is in base-devel
[13:06:26] <abaumann> this doesn't make sense
[13:06:31] <deep42thought> so it's a dependency of everything
[13:06:38] <deep42thought> a makedepends
[13:06:45] <abaumann> so, util-linux has to be rebuilt first.
[13:06:54] <abaumann> and blocks all the rest?
[13:07:06] <deep42thought> the question is, why the buildmaster does not see the loop
[13:07:12] <deep42thought> oh, wait
[13:07:19] <deep42thought> maybe I ignore base and base-devel
[13:07:38] <abaumann> there are some tiny loops in base and base-devel ;-)
[13:07:45] <deep42thought> exactly
[13:09:06] <deep42thought> *grrr* stupid typo
[13:09:24] <abaumann> hum?
[13:09:43] <deep42thought> `dependency_types`.`relevant_for_building` OR `dependency_types`.`relevant_for_building`
[13:09:58] <deep42thought> one of both should be `dependency_types`.`relevant_for_binary_packages`
[13:10:10] <deep42thought> although ...
[13:10:12] <deep42thought> hmm
[13:10:16] <deep42thought> shouldn't make a difference
[13:12:34] -!- bulletmark has quit [Ping timeout: 240 seconds]
[13:13:42] <abaumann> just testing if chromium is building now locally with fatal_linker_warnings=false
[13:13:47] <deep42thought> hmm, it _does_ see the loop
[13:13:48] <deep42thought> strange
[13:46:08] <abaumann> ah. util-linux has built, so some other packages are building now. the ones with highest prio first, like llvm.
[13:46:16] <deep42thought> \o/
[13:46:25] <abaumann> so everything is ok with the priority system, just the look in base was blocking things.
[13:46:28] <abaumann> :-)
[13:46:43] <deep42thought> get-package-updates currently hickups on go
[13:46:48] <deep42thought> which moved from extra to community
[13:46:53] <deep42thought> I'm looking into this right now
[13:47:08] <abaumann> ok.
[13:47:18] <abaumann> I'm fixing some packages and I'm doing the audio build on i486.
[13:59:21] -!- ofara has joined #archlinux32
[14:18:23] -!- ofara has quit [Ping timeout: 245 seconds]
[14:34:15] <deep42thought> looks like I sucessfully repaired the moving from [extra] to [community] :-)
[14:34:33] -!- oaken-source has quit [Ping timeout: 252 seconds]
[14:46:09] <abaumann> jippie! packages ahoi. :-)
[14:46:26] -!- Alina-malina has joined #archlinux32
[14:47:29] <deep42thought> todo#41 and #88 are still todos (and quite important)
[14:47:30] <deep42thought> :-/
[14:51:55] <deep42thought> btw: I plan to start signing our package database some time in the future
[14:52:16] <abaumann> with a second key?
[14:52:20] <deep42thought> yes
[14:52:26] <deep42thought> which will be only on the build master
[14:52:41] <abaumann> that makes sense.
[14:53:10] <deep42thought> there is already some necessary modification in pacman.conf: SigLevel = Required DatabaseOptional DatabaseTrustAll
[14:53:37] <deep42thought> this way, it won't break when the buildmaster key is not in the pacman keyring
[14:53:47] <deep42thought> because I do not intend to put it into the keyring
[14:53:56] <deep42thought> because it should not be allowed to sign packages
[14:54:05] <deep42thought> because otherwise, the buildmaster could corrupt packages
[14:56:14] -!- bulletmark has joined #archlinux32
[14:57:02] <abaumann> I reckon upstream does it in the same way?
[14:58:31] -!- bulletmark has quit [Remote host closed the connection]
[14:58:37] <deep42thought> no
[14:58:44] <deep42thought> they don't sign the db at all
[14:59:07] <deep42thought> and the default options in pacman.conf prevent signing the database iff the signing key is not also in the keyring
[14:59:30] <deep42thought> e.g. "DatabaseTrustAll" is additional in our config
[14:59:56] <abaumann> mmh. they don't sign it. they don't have a central build master which could sign it?
[15:01:22] <deep42thought> well, they have a central server (like our master mirror) where they could sign it
[15:04:14] <deep42thought> the question is: what do you gain if you do not enforce that the db is signed with a trusted key
[15:04:24] <deep42thought> and apparently they decided, that it's not worth the effort
[15:04:54] <deep42thought> but I think, we should sign it and make it verifyable - at least manually
[15:26:08] <abaumann> oh.. my build slave crashed..
[15:31:54] -!- ofara has joined #archlinux32
[15:33:06] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.stdin.2018-09-01T15:24:36.2T82XU"?.
[15:33:42] <deep42thought> umpf
[15:34:20] <deep42thought> I have a huge manual merge running currently :-/
[15:34:24] <deep42thought> s/merge/join/
[15:40:50] <abaumann> big join? because of sanity checking?
[15:41:27] <deep42thought> no
[15:41:37] <deep42thought> because of putting all versions into a single table
[15:41:58] <deep42thought> and because I forgot a "UNIQUE KEY", they were multiple times in that table
[15:42:04] <deep42thought> (avg. 4 times each)
[15:42:16] <deep42thought> and the merge of that table with binary_packages was just too much
[15:45:07] <abaumann> ah
[15:55:23] -!- morealaz has joined #archlinux32
[16:16:34] -!- morealaz has quit [Remote host closed the connection]
[16:47:33] <abaumann> mmh. I have a mpd-light for i486, let's test on the Alix. :-)
[16:53:16] <abaumann> (10/35) upgrading p11-kit [##################################] 100%
[16:53:19] <abaumann> error: command terminated by signal 4: Illegal instruction
[16:53:21] <abaumann> *grmpf*
[16:53:28] <abaumann> some SSE code slipped in
[16:54:50] <abaumann> (26/35) upgrading iputils [##################################] 100%
[16:54:53] <abaumann> error: command terminated by signal 4: Illegal instruction
[16:54:56] <abaumann> (28/35) upgrading lvm2 [##################################] 100%
[16:54:59] <abaumann> error: command terminated by signal 4: Illegal instruction
[16:55:00] <abaumann> (30/35) upgrading man-pages [###################---------------] 57%
[16:55:03] <abaumann> error: command terminated by signal 4: Illegal instruction
[16:55:25] <abaumann> This is really annoying when you tell a compiler to produce code for an i486 and the the compiler goes away and does something beatifully different
[16:55:58] <abaumann> ok. bricked the alix (again) :-(
[16:56:46] <abaumann> so, this means the whole 8.2 toolchain is borked and also the complete rebuild for i486.
[17:00:59] <abaumann> what really annoys me. qemu 486 emulation doesn't detect this kind of illegal instructions..
[17:02:06] <abaumann> I think, I have to look at qemu-system-i386 and see, how this emulation is really made..
[17:08:50] <abaumann> ah well. another day.. :-)
[17:08:51] -!- abaumann has quit [Quit: leaving]
[17:11:36] -!- morealaz has joined #archlinux32
[17:48:49] * buildmaster resumes sanity.
[18:12:41] -!- morealaz has quit [Remote host closed the connection]
[18:32:44] <buildmaster> go-tools is broken (says buildknecht).
[18:54:12] <buildmaster> firefox-developer-edition is broken (says rechenknecht).
[19:36:40] -!- oaken-source has joined #archlinux32
[19:56:31] -!- lukeshu has quit [*.net *.split]
[19:56:31] -!- prurigro has quit [*.net *.split]
[19:59:18] -!- lukeshu has joined #archlinux32
[20:01:49] -!- prurigro has joined #archlinux32
[20:35:18] -!- woshty has joined #archlinux32
[20:41:29] -!- woshty has quit [Ping timeout: 244 seconds]
[20:42:24] <buildmaster> influxdb is broken (says buildknecht).
[20:49:43] -!- woshty has joined #archlinux32
[21:18:07] -!- z3ntu has quit [Remote host closed the connection]
[21:24:48] <deep42thought> tyzoid: Have you seen this: https://bbs.archlinux32.org ?
[21:24:50] <phrik> Title:Gitea on git.archlinux32.org could do with an update / Creating/Maintaining Packages / Arch Linux 32 Forums (at bbs.archlinux32.org)
[21:33:47] -!- z3ntu has joined #archlinux32
[22:11:43] -!- oaken-source has quit [Ping timeout: 245 seconds]
[22:26:55] <buildmaster> electron is broken (says nlopc46).
[22:48:05] -!- deep42thought has quit [Quit: Leaving.]
[23:39:33] -!- finsternis has joined #archlinux32