#archlinux32 | Logs for 2024-01-18

Back
[00:45:52] -!- epony has quit [Remote host closed the connection]
[00:47:40] -!- epony has joined #archlinux32
[01:02:37] -!- epony has quit [Remote host closed the connection]
[01:03:57] -!- epony has joined #archlinux32
[01:16:24] mavicaway is now known as mavica
[04:09:04] -!- bill-auger has quit [Quit: No Ping reply in 210 seconds.]
[04:10:33] -!- bill-auger has joined #archlinux32
[04:34:40] -!- epony has quit [Remote host closed the connection]
[04:36:23] -!- epony has joined #archlinux32
[05:01:49] -!- buildmaster has quit [Read error: Connection reset by peer]
[05:04:12] -!- buildmaster has joined #archlinux32
[06:16:01] mavica is now known as mavicaway
[08:10:02] -!- abaumann has joined #archlinux32
[08:10:02] <buildmaster> Hi abaumann!
[08:10:02] <buildmaster> !rq abaumann
[08:10:03] <phrik> buildmaster: <abaumann> OOM-killer could order memory on ebay..
[08:10:42] <abaumann> KitsuWhooa: I'll have a look at copr-cli. The problem with the state git repo is, that nothing is checking whether it points to a valid git hash (I personally would do some serious git hook checking there).
[08:11:05] <abaumann> But it could also be something else, I'll have a look at the pkginfo.php script first, because that is the most likely culprit. :-)
[08:21:05] <KitsuWhooa> Nahh
[08:21:12] <KitsuWhooa> Actually maybe
[08:21:22] <KitsuWhooa> When I put it in gitlab I think it brings something up
[08:21:36] <KitsuWhooa> But if you look at the commit history the hash is nowhere to be found
[08:22:18] <KitsuWhooa> As a sidenote, say I want to open a merge request or equivalent for changes to the builder scripts
[08:22:24] <KitsuWhooa> Is there a process for that?
[08:22:48] <KitsuWhooa> I made schedule for rebuild ask you which hash you want to build because it always picked the last one and it was wrong
[08:22:59] <KitsuWhooa> And I thought it might be useful to have in general
[08:30:04] <abaumann> You can send me a 'git patch', or we just add you to the git repo writers (but I have to find out how that works exactly)
[08:30:07] <abaumann> cat 'python-wheel=community=eb54f1e97f245cab505646f1eae8022bd8771be1=0000000000000000000000000000000000000000'
[08:30:10] <abaumann> curl: (22) The requested URL returned error: 500
[08:30:10] <abaumann> is another one I found
[08:30:19] <abaumann> Maybe the git hashes are pointing to tags
[08:30:42] <abaumann> git clone https://gitlab.archlinux.org
[08:30:44] <phrik> Title: Arch Linux / Packaging / State · GitLab (at gitlab.archlinux.org)
[08:30:46] <abaumann> cat state/core-x86_64/which
[08:30:46] <abaumann> which 2.21-6 2.21-6 be0a557573b9fe210e37e85bc69f19a4c7ec31ea
[08:30:50] <abaumann> git -C which show-ref --tags
[08:31:01] <abaumann> be0a557573b9fe210e37e85bc69f19a4c7ec31ea refs/tags/2.21-6
[08:31:14] <abaumann> that's the usualy way the state repo points to the local gitlab repo of the package
[08:31:29] <abaumann> pkgctl repo clone --protocol=https --arch32 which
[08:31:48] <abaumann> with a devtools32 package installed should also work to fetch the package (already patched for arch32 builds)
[08:32:28] <abaumann> I have once again to merge devtools upstream, deep42thoughts and my changed to it in the old devtools into devtools32
[08:33:05] <KitsuWhooa> Mhm
[08:33:33] <abaumann> didn't I give you write access to the git repo already?
[08:34:00] <abaumann> I see a tasos.pub in gitolite-admin/keydir
[08:34:29] <abaumann> @admins = abaumann .. tasos
[08:34:41] <abaumann> so in theory you should be able to push changes via ssh
[08:35:03] <abaumann> (minus obvious configuration bugs on my side, of course :-) )
[08:35:10] <KitsuWhooa> Ooooh
[08:35:23] <KitsuWhooa> I was asking more to get feedback / review
[08:35:31] <abaumann> oh.
[08:35:41] <abaumann> well.
[08:35:53] <KitsuWhooa> But I also didn't know I can push to it
[08:35:58] <KitsuWhooa> I thought I only had access to packages
[08:36:03] <abaumann> the process was always: change it, if it breaks, fix it or revert it and clean up the mess you produced.
[08:36:05] <KitsuWhooa> Now I know, thanks!
[08:36:06] <abaumann> :-)
[08:36:08] <KitsuWhooa> I see
[08:36:11] <KitsuWhooa> Sounds good to me
[08:36:38] <KitsuWhooa> If you're curious https://tasossah.com
[08:36:40] <phrik> Title: schedule-for-rebuild diff (at tasossah.com)
[08:36:45] <KitsuWhooa> This isn't a proper patch obviously
[08:36:49] <abaumann> the mess produced part usually gets fixed by deep42thought.. who fixes it calmly without complaining (in constract to myself) :-)
[08:37:15] <KitsuWhooa> Hahaha
[08:37:19] <KitsuWhooa> I like to complain too
[08:37:21] <abaumann> this is not a core package I would say, so adding things to it is not super critical.,
[08:37:54] <abaumann> your sorting the things, ok.
[08:38:28] <abaumann> ah. this is a sort unique
[08:38:34] <KitsuWhooa> Yeah
[08:38:38] <abaumann> that shouldn't do much harm, or does it?
[08:38:46] <KitsuWhooa> It was sorting on its own
[08:38:53] <KitsuWhooa> As in, I didn't add the sort
[08:39:05] <KitsuWhooa> There's also another sort above
[08:39:49] <KitsuWhooa> I don't know why it breaks, but as it is right now, it only seems to pick the last package
[08:40:20] <KitsuWhooa> When there are multiple commits for the same package, it picks the last one, and sometimes it's outdated or outright no longer exists
[08:40:26] <KitsuWhooa> And with this I can work around it
[08:40:35] <KitsuWhooa> It's how I got a few more stuck packages to build
[08:40:55] <KitsuWhooa> And by stuck I mean I'd run the script, it'd show all the hashes, and it'd never build it
[08:41:52] <KitsuWhooa> I suspect the root cause is something else, but I couldn't figure it out, so for a workaround this is fine I think
[08:42:04] <abaumann> +echo "Which packages to rebuild? (Comma separated; Empty means all)"
[08:42:04] <abaumann> +nl "${tmp_dir}/pkgbases"
[08:42:04] <abaumann> +read pkg_rebuild_choice
[08:42:12] <abaumann> making it interactive is not the best idea maybe?
[08:42:18] <abaumann> bin/return-assignment: call schedule-for-rebuild instead of seed-build-list
[08:42:24] <abaumann> this script is called by other scripts.
[08:42:34] <KitsuWhooa> Yeah. One of the things I want to do is require a flag to be interactive
[08:42:39] <KitsuWhooa> Precisely for this reason
[08:42:39] <abaumann> exactly
[08:42:47] <abaumann> then it's fine
[08:42:52] <KitsuWhooa> Awesome
[08:43:23] <abaumann> remember: I also don't really know much about the buildmaster.. so reviews are - aeh - partial :-)
[08:43:50] <KitsuWhooa> I know, yeah. But two pairs of eyes are better than nothing :p
[08:44:01] <abaumann> true :-)
[08:44:29] <KitsuWhooa> Er
[08:44:34] <KitsuWhooa> I meant "than one"
[08:46:35] <abaumann> mmh, I see some 40 packages with broken pkginfo files on the buildmaster, most of them are python-XX somehting.
[08:46:56] <abaumann> they might point to weird places, so we need the pkginfo.php fix in the end
[08:48:37] <abaumann> cat state/extra-any/python-wheel
[08:48:37] <abaumann> python-wheel 0.42.0-1 0.42.0-1 5086674b1b412ed4f9eedc3c6c0629fec3acdd0b
[08:48:40] <abaumann> git -C python-wheel show-ref --tags | grep 5086674b1b412ed4f9eedc3c6c0629fec3acdd0b
[08:48:43] <KitsuWhooa> Yeah
[08:48:45] <abaumann> 5086674b1b412ed4f9eedc3c6c0629fec3acdd0b refs/tags/0.42.0-1
[08:48:48] <KitsuWhooa> I don't know why they exist
[08:48:51] <abaumann> this looks ok
[08:49:03] <KitsuWhooa> I have a query I run to get the highest priority packages in the build list
[08:49:12] <KitsuWhooa> And there are quite a few just sitting at the top, broken
[08:49:26] <abaumann> cat python-wheel\=extra\=5086674b1b412ed4f9eedc3c6c0629fec3acdd0b\=0000000000000000000000000000000000000000
[08:49:29] <abaumann> is ok
[08:49:37] <abaumann> 'python-wheel=community=7dbe255b44d62a31650a9f1a149b0447504b4943=0000000000000000000000000000000000000000'
[08:49:40] <abaumann> 'python-wheel=community=eb54f1e97f245cab505646f1eae8022bd8771be1=0000000000000000000000000000000000000000'
[08:49:47] <KitsuWhooa> Those aren't :p
[08:49:49] <abaumann> those are rather bizare, as there is no such thing as community anymore
[08:49:54] <KitsuWhooa> Yes
[08:50:08] <KitsuWhooa> I think it might be because I used -f on schedule for rebuild
[08:50:32] <KitsuWhooa> I had to use it because otherwise it'd see a community package, the php script would error, and it wouldn't proceed any further
[08:50:53] <KitsuWhooa> If you know/figure out how to clean this stuff up, please let me know
[08:51:05] <abaumann> so, I have to implement the fall back into gitsvn and community in this case
[08:51:13] <abaumann> rm *
[08:51:16] <abaumann> ;-)
[08:51:21] <KitsuWhooa> But it'd build them anyway
[08:51:25] <KitsuWhooa> No?
[08:51:35] <KitsuWhooa> Won't that succeeding build ancient packages?
[08:52:02] <abaumann> yes, but maybe the buildmaster has to build ancient packages because of some depenendencies?
[08:52:24] <KitsuWhooa> Hm
[08:52:26] <KitsuWhooa> Perhaps
[08:52:40] <KitsuWhooa> Although the python packages I managed to get built still have at least two entries
[08:52:44] <KitsuWhooa> One community and one extra
[08:52:51] <abaumann> For sure the pkginfo.php caching mechanism only works for the new git/state gitlab repo case without community
[08:52:56] <KitsuWhooa> Yeah
[08:56:02] <abaumann> ./get-source-info python-wheel community eb54f1e97f245cab505646f1eae8022bd8771be1 0000000000000000000000000000000000000000
[08:56:05] <abaumann> curl: (22) The requested URL returned error: 500
[08:56:16] <abaumann> without -c it doesn't run via pkginfo.php on the buildmaster
[08:56:26] <abaumann> so the problem is in the core logic, not in the caching mechanism
[08:57:19] <abaumann> I fear, it's the upstream package fetcher
[08:57:40] <abaumann> curl -o /home/master/builder/work/tmp.make_source_info.1R0CKv/tmp.upstream-package.UrNI3VIM7J -fLSs https://buildmaster.archlinux32.org
[08:57:44] <abaumann> curl: (22) The requested URL returned error: 500
[08:58:11] <abaumann> that one always goes via the upstream cache on the buildmaster (because of the bandwith limitations upstream)
[08:59:12] <abaumann> python-wheel-3c882b9c66f2e478092cf5b35cca6ba9726347fa.tar.gz and python-wheel-5086674b1b412ed4f9eedc3c6c0629fec3acdd0b.tar.gz
[08:59:15] <abaumann> python-wheel 0.42.0-1 0.42.0-1 5086674b1b412ed4f9eedc3c6c0629fec3acdd0b
[08:59:18] <abaumann> 3c882b9c66f2e478092cf5b35cca6ba9726347fa refs/tags/0.40.0-3
[08:59:24] <abaumann> those are two new upstream revisions
[09:00:18] <abaumann> yep: /srv/http/upstream-packages.php only fetches new packages from upstream gitlab https://gitlab.archlinux.org
[09:00:19] <phrik> Title: Packages · GitLab (at gitlab.archlinux.org)
[09:00:36] <abaumann> This is something I botched together in Summer 2023..
[09:27:53] -!- epony has quit [Remote host closed the connection]
[09:28:52] <KitsuWhooa> Works pretty well for what it is honestly
[09:28:54] -!- epony has joined #archlinux32
[11:29:52] -!- bill-auger has quit [Ping timeout: 255 seconds]
[11:36:11] -!- bill-auger has joined #archlinux32
[11:56:39] -!- bill-auger has quit [Remote host closed the connection]
[11:57:04] -!- bill-auger has joined #archlinux32
[13:32:19] <abaumann> mmh, strange, just removing the pkginfo file for copr-cli, and now get-package-updates works again..
[13:32:31] <abaumann> I made a non-gui, non-emacs, non-docu patch for cmake
[13:57:28] <KitsuWhooa> I really didn't want to go that route, which is why I've been fighting python
[13:57:35] <KitsuWhooa> but oh well
[14:00:05] <abaumann> just temporary :-)
[14:00:25] <KitsuWhooa> yeah
[14:00:36] <abaumann> cmake will also have a hard time to build because of qt6 and especially python-spinx
[14:00:47] <abaumann> even if python is fixed, python-spinx will not work
[14:00:55] <KitsuWhooa> I'm sure I can get it to :p
[14:01:03] <abaumann> :-)
[14:04:55] <KitsuWhooa> is there a specific reason you're saying it won't work?
[14:07:25] <abaumann> it didn't in the past. :-)
[14:07:39] <KitsuWhooa> ah!
[14:07:52] <abaumann> I don't recall which parts didn't build and just didn't have the patience to dig deeper.
[14:07:55] <abaumann> :-)
[14:07:55] <KitsuWhooa> fair
[14:08:07] <KitsuWhooa> I'm wondering how easy it'd be to add a new straw that is build support without check
[14:08:46] <KitsuWhooa> I think it'd be useful
[14:08:56] <abaumann> Something just doing a 'unset check' and an 'unset checkdepends', yes, quite helpful.
[14:09:11] <KitsuWhooa> yeah, in an automated way
[14:09:19] <abaumann> Especially because python now sucks in tons of dependencies which are only needed for testing..
[14:09:23] <KitsuWhooa> yup
[14:09:33] <KitsuWhooa> although I think the upgrade to 3.12 will be easier
[14:09:41] <abaumann> The same goes for documentation, but there a generic hook is not so easy to do..
[14:09:48] <KitsuWhooa> one of the main issues I had was that the dependencies for the existing packages are wrong
[14:11:28] <KitsuWhooa> for example things depend on jaraco.text which depends on jaraco.functools
[14:11:34] <KitsuWhooa> but I can't yet build jaraco.text
[14:11:42] <KitsuWhooa> and packages that depend on it fail with no module functools :p
[14:12:08] <KitsuWhooa> because pacman doesn't install it as the current jaraco.text doesn't bring it
[14:12:15] <KitsuWhooa> *bring it in
[14:12:20] <abaumann> I'm not sharing your optimism that 3.12 will be better.
[14:12:34] <abaumann> then more of old bootstrapping stuff will have been forgotten or deprectated.
[14:12:39] <KitsuWhooa> I'll let you know when I get to it :p
[14:12:45] <abaumann> python takes a strange route currently IMHO
[14:12:53] <KitsuWhooa> I just hope it happens after I get 3.11 up and running properly
[14:12:59] <abaumann> hey, get-package-updates passed :-)
[14:13:03] <KitsuWhooa> awesome!
[14:13:06] <KitsuWhooa> thanks for fixing it
[14:13:12] <abaumann> np
[14:40:09] <abaumann> ./schedule-for-rebuild -w -j -p '^cmake$'
[14:40:09] <abaumann> cmake 100aa9c3ed3e2469b6919caf16c0112cda206faf 7755779c2876126e2fef716b7988300f14eb1a1f 1a340014bef8f691043c48dc34502afdc913f765 extra
[14:40:12] <abaumann> "make_source_info" did not return a "pkgbase" - eh, what?
[14:40:15] <abaumann> >curl: (22) The requested URL returned error: 500<
[14:40:17] <abaumann> ><
[14:40:20] <abaumann> ><
[14:40:22] <abaumann> mmh.
[14:40:46] <abaumann> now we have three hashes there..
[14:41:43] <abaumann> more cmake\=extra\=100aa9c3ed3e2469b6919caf16c0112cda206faf\=7755779c2876126e2fef716b7988300f14eb1a1f
[14:41:46] <abaumann> curl: (22) The requested URL returned error: 500
[14:41:50] <abaumann> aha.
[14:43:11] <abaumann> now cmake should have a new hash and a now modified revision from arch32, so it fails? well, more digging for me. :-)
[14:43:32] <KitsuWhooa> use my script and pick the correct one ;)
[14:44:21] <KitsuWhooa> schedule-for-rebuild-test-no-sort -w -j -p '^cmake$'
[14:44:47] <abaumann> oh. that's what it is for :-)
[14:44:50] <KitsuWhooa> yup
[14:44:50] <abaumann> now I understand
[14:44:58] <KitsuWhooa> it'll show you the hashes and you get to pick which ones you want
[14:45:22] <KitsuWhooa> it won't attempt to run make_source_info on the community one if you don't select it and it won't error out
[14:45:30] <KitsuWhooa> and if there are multiple ones in extra, you can pick the right one
[14:45:45] <abaumann> so, I just have to _find_ the right one, ok :-)
[14:46:21] <KitsuWhooa> I could make it print ready to click gitlab links...
[14:46:40] <KitsuWhooa> https://gitlab.archlinux.org{package}/-/commit/${commit}
[14:46:42] <phrik> Title: Sign in · GitLab (at gitlab.archlinux.org)
[14:46:46] <KitsuWhooa> something like that
[14:46:47] <abaumann> I fear, something is wrong deeper down, because this script is supposed to be called by others
[14:46:51] <KitsuWhooa> yes
[14:46:58] <KitsuWhooa> it absolutely is
[14:47:06] <KitsuWhooa> but it's a temporary workaround to get things going when they get stuck
[14:47:16] <abaumann> at least for now we can force the right versions to build, ok.
[14:47:20] <KitsuWhooa> I suspect there shouldn't be multiple commits for the same repository (i.e. extra)
[14:47:22] <KitsuWhooa> yup
[14:47:45] <abaumann> mmh, probably not..
[14:48:11] <KitsuWhooa> I also had to create a PKGBUILD for a few packages that buildmaster was convinced there was a mod hash/commit
[14:48:17] <abaumann> d4b9dbd1537abd94c516dd0de5e283f6bf5b84dd aha.
[14:48:22] <abaumann> cmake d4b9dbd1537abd94c516dd0de5e283f6bf5b84dd 1a340014bef8f691043c48dc34502afdc913f765 1a340014bef8f691043c48dc34502afdc913f765 extra
[14:48:23] <KitsuWhooa> there probably was in the past, but it was deleted at some point and buildmaster doesn't seem to have realised
[14:48:25] <abaumann> ok.
[14:48:42] <abaumann> yep, that went fine.
[14:48:49] <KitsuWhooa> 👍
[14:48:55] <abaumann> 1 any cmake 100aa9c3ed3e2469b6919caf16c0112cda206faf 7755779c2876126e2fef716b7988300f14eb1a1f extra
[14:48:58] <abaumann> 2 any cmake 2bc74eea5f2c582f35d7019e156ce50550898b01 43ab1bde884297633fcfc9f476d517e616763a72 extra
[14:49:01] <abaumann> 3 any cmake 306cd8bb0f983b7a2596fb252e3ad014323853fe 01872969762e2c7cbaa8dfa03cb2b2f565a8625b extra
[14:49:04] <abaumann> 4 any cmake 499e2eef992bb28010f406d4fb67d87da72b1ddb 818c917c48f8c092a05043bd3cf34b68e3feae7f extra
[14:49:07] <abaumann> 5 any cmake d4b9dbd1537abd94c516dd0de5e283f6bf5b84dd 1a340014bef8f691043c48dc34502afdc913f765 extra
[14:49:10] <KitsuWhooa> ...oh god
[14:49:10] <abaumann> this sounds quite wrong though.
[14:49:16] <KitsuWhooa> I've never seen this many
[14:49:18] <KitsuWhooa> usually it's 3 at most
[14:49:29] <abaumann> I fear, it's accumulating for some reason..
[14:49:32] <KitsuWhooa> it is
[14:49:49] <KitsuWhooa> and I don't know how to clean them up
[14:49:57] <KitsuWhooa> I'm still afraid to touch the database
[14:50:12] <abaumann> 1a340014bef8f691043c48dc34502afdc913f765 is the last commit in arch32 packages, that's also the right mod_revision.
[14:50:18] <abaumann> ok, so the right version of cmake is built.
[14:50:26] <abaumann> yes, me too.
[14:50:37] <KitsuWhooa> speaking of the database
[14:50:38] <abaumann> let's see, maybe if a package builds correctly, the cleanup happens.
[14:50:39] <KitsuWhooa> SELECT * FROM build_assignments INNER JOIN package_sources ON build_assignments.package_source = package_sources.id ORDER BY build_assignments.priority DESC LIMIT 10;
[14:50:51] <KitsuWhooa> I use this to see what is queued up to be built
[14:51:19] <KitsuWhooa> build-tasossah-3 is building cmake
[14:51:22] <KitsuWhooa> so that worked
[14:51:35] <KitsuWhooa> and so is eurobuild6-5
[14:51:53] <abaumann> yep. I see it on https://buildmaster-status.archlinux32.org
[14:51:56] <abaumann> fine.
[14:52:14] <abaumann> i486 is a special case, I don't even know, if those VMs of mine are still ok.. have to check..
[14:52:21] <KitsuWhooa> yeah...
[14:53:00] <abaumann> mmh. schedule-for-rebuild should actually sort by latest revision then latest mod_revision
[14:53:08] <KitsuWhooa> it doesn't :p
[14:53:21] <abaumann> apparently :-)
[14:53:27] <KitsuWhooa> found out the hard way
[14:53:34] <abaumann> can imagine. sorry about that.
[15:00:56] <KitsuWhooa> no worries
[15:01:56] <abaumann> If I have a look at the SQL and the shell code IMHO this has to be cleaned up heavily before even trying to fix anything. 100nds of lines composing a SQL statement with printf also calling join functions etc. sounds like hard to maintain.
[15:02:44] <abaumann> I don't see a sort there in the database, I see a DISTINCT on top, so why a sort -u below to remove duplicates.
[15:02:48] <abaumann> and all this kind of things.
[15:02:57] <abaumann> touching it and making one error corrupts the database.
[15:02:59] <KitsuWhooa> I do not understand that either
[15:03:01] <abaumann> so yeah.
[15:03:08] <KitsuWhooa> are there automated database backups?
[15:03:17] <abaumann> yes, there are.
[15:03:26] <KitsuWhooa> that's something at least
[15:03:27] <abaumann> but: they might not fit to the mirror on the buildmaster then.
[15:03:42] <abaumann> so this requires a huge synchronization effort if things go wrong.
[15:03:57] <KitsuWhooa> Hm
[15:04:12] <abaumann> the mirror is also backuped, but not in snapshots synchronous to the database dumps
[15:04:24] <KitsuWhooa> yeah, we could synchronise these two
[15:04:28] <KitsuWhooa> mysqldump and then rsync
[15:04:36] <abaumann> cloning the thing to generate a test instance to play with is also a hard task
[15:04:43] <KitsuWhooa> indeed
[15:05:26] <abaumann> the tactics to write some new tools around the problems, like you did, is a viable way.
[15:05:55] <abaumann> I would go even further, write tools which can enter a package built outside the buildmaster and register it somehow in the database and the mirror.
[15:06:12] <abaumann> then slowly work around and build packages outside when they require patching
[15:06:27] <abaumann> but for that they have to work correctly
[15:07:08] <KitsuWhooa> mhm
[15:11:20] <KitsuWhooa> I should make a few more builders
[15:14:59] <abaumann> ./schedule-for-rebuild-test-no-sort -w -j -p '^filesystem$'
[15:14:59] <abaumann> Which packages to rebuild? (Comma separated; Empty means all) 1 any filesystem 073e0a7beb72969fd619b32618c7b1667fb37597 b721e73387f9daad6844cc84e46396411b77bdfd core 2 any filesystem 9568bc24b9c8da5234ef7b95aaaaa631e0bc311b b721e73387f9daad6844cc84e46396411b77bdfd core
[15:15:07] -!- drathir_tor has quit [Remote host closed the connection]
[15:15:19] <abaumann> aha. this also explains why we had an old filesystem package show up in core.
[15:15:36] <KitsuWhooa> for what it's worth, ~/builder/bin is in PATH
[15:15:40] <KitsuWhooa> so you don't need to ./ it
[15:15:58] -!- drathir_tor has joined #archlinux32
[15:16:02] <abaumann> filesystem 2023.09.18-1 2023.09.18-1 073e0a7beb72969fd619b32618c7b1667fb37597
[15:16:06] <abaumann> this is the right one
[15:16:15] <KitsuWhooa> and it's the first one
[15:16:17] <abaumann> let's force it.
[15:16:18] <abaumann> yep.
[15:16:19] <KitsuWhooa> and because it was the first one, it wouldn't get built
[15:16:25] <KitsuWhooa> it'd only pick the last one
[15:16:35] <abaumann> 073e0a7beb72969fd619b32618c7b1667fb37597 is the upstream commit
[15:19:42] <abaumann> /usr/share/webapps/fluxbb/cache/ drwxrwxr-x 2 http http 4096 Feb 29 2020 .
[15:19:47] <abaumann> and php-fpm runs as http
[15:19:48] <abaumann> nice
[15:20:15] <abaumann> this darn spam in the bbs..
[15:20:54] <KitsuWhooa> I wonder if manual approval would make things better
[15:21:01] <abaumann> I start to understand why upstream has no registration to gitlab and makes you mail in if you need an account.
[15:21:04] <abaumann> exactly.
[15:21:08] <KitsuWhooa> yes
[15:21:19] <KitsuWhooa> although we could probably have a register form and a "why do you want to register"
[15:21:25] <KitsuWhooa> so we can quickly see someone saying "I want to report an issue"
[15:21:30] <KitsuWhooa> "ask advice for my old laptop"
[15:21:33] <KitsuWhooa> and quickly approve it
[15:21:41] <KitsuWhooa> but I assume someone needs to program that :p
[15:21:49] -!- drathir_tor has quit [Remote host closed the connection]
[15:21:56] <abaumann> I think they have to solve 3 capchas already now if they register.
[15:22:15] <abaumann> Obviously there are people being paid just to do that and then to use the accounts to spam.
[15:22:17] <KitsuWhooa> the problem with captchas is that there are services that you pay x amount of money and they solve it for you
[15:22:18] -!- drathir_tor has joined #archlinux32
[15:22:20] <abaumann> Sad new world..
[15:22:23] <KitsuWhooa> mhm
[15:23:05] <abaumann> Yes. Modern world where people solve puzzles in the home office in third world countries or train artificial intelligence by clickig on "is it a cat" or "is it a camel".
[15:23:25] <abaumann> this is unethical at best.
[15:23:56] <KitsuWhooa> I don't think they are concerned about it sadly
[15:23:58] <abaumann> so, we could argue that by adding captchas we make the world a even sader place.. something to ponder.
[15:24:22] <KitsuWhooa> we could also have a list of questions
[15:24:26] <KitsuWhooa> but they eventually figure those out too
[15:24:40] <KitsuWhooa> especially if they have AI trying random answers
[15:24:52] <KitsuWhooa> "Which is one of the supported architectures?"
[15:24:54] <abaumann> I wanted to clean the bbs and the bug reporting tool. Maybe remove all accounts which don't have a post or comment or bug at all?
[15:25:00] <KitsuWhooa> might be vague enough
[15:25:02] <abaumann> but it's work and maybe too agressive.
[15:25:07] <KitsuWhooa> agreed
[15:25:31] <abaumann> "Which is one of the supported architectures?"
[15:25:39] <abaumann> hah, can I even answer that? ;-)
[15:26:21] <KitsuWhooa> or "what's the output of pacman -whatever" ;p
[15:26:26] <KitsuWhooa> I think upstream does that for the wiki
[15:26:30] <abaumann> true
[15:26:43] <KitsuWhooa> although that might be to discourage non arch users from editing it
[15:26:47] <KitsuWhooa> not sure
[15:27:19] <abaumann> spam is not that dramatic.. not anymore.. maybe also because more and more people move away from arch32
[15:27:28] <KitsuWhooa> I don't think so
[15:27:29] <abaumann> it's hard to tell how big the user basis is..
[15:27:44] <KitsuWhooa> spambots just scour the internet
[15:28:04] <abaumann> you're probably right
[16:05:08] -!- abaumann has quit [Quit: leaving]
[16:18:06] -!- epony has quit [Quit: QUIT]
[17:06:31] -!- drathir_tor has quit [Ping timeout: 240 seconds]
[17:16:11] -!- drathir_tor has joined #archlinux32
[17:18:22] <KitsuWhooa> RIP buildmaster went insane
[17:20:02] -!- epony has joined #archlinux32
[17:47:32] -!- abaumann has joined #archlinux32
[17:47:33] <buildmaster> Hi abaumann!
[17:47:33] <buildmaster> !rq abaumann
[17:47:33] <phrik> buildmaster: <abaumann> works. I receive my personal spam now ;-)
[17:47:49] <abaumann> for build-list/deletion list corruption I use the following:
[17:47:56] <abaumann> mysql buildmaster -e 'delete b_bpir from binary_packages as a join binary_packages as b on a.pkgname=b.pkgname and a.architecture=b.architecture'"$(mysql_join_binary_packages_binary_packages_in_repositories a a_bpir)$(mysql_join_binary_packages_binary_packages_in_repositories b b_bpir)$(mysql_join_binary_packages_in_repositories_repositories b_bpir b_r)"' where a_bpir.repository=9 and
[17:48:02] <abaumann> b_r.name="deletion-list"'
[17:48:25] <abaumann> or used. it fails now?
[17:50:07] <abaumann> it doesn't do the right thing or nothing at all, the error persists.
[17:51:13] <abaumann> this is so annoying.
[17:51:47] <abaumann> ERROR 1109 (42S02): Unknown table 'b_bpir' in MULTI DELETE
[17:51:51] <abaumann> aha. since when?
[17:52:43] <abaumann> no sorry, cannot fix that.
[17:52:44] -!- abaumann has quit [Quit: leaving]
[18:06:01] <KitsuWhooa> I'll look into it
[18:18:59] mavicaway is now known as mavica
[19:42:07] <KitsuWhooa> ugh, this is pretty bad
[20:19:28] mavica is now known as mavicaway
[20:20:34] mavicaway is now known as mavica
[21:20:57] mavica is now known as mavicaway
[21:22:01] mavicaway is now known as mavica
[23:00:12] -!- epony has quit [Remote host closed the connection]
[23:15:17] -!- T`aZ has quit [Ping timeout: 240 seconds]
[23:21:12] -!- epony has joined #archlinux32
[23:39:32] -!- T`aZ has joined #archlinux32