2023-01-02 06:23:28 <ikke> psykose_: stopped the bot for now
2023-01-02 06:23:37 <psykose_> okie
2023-01-02 06:38:11 <ikke> psykose_: I've started it again with autolabeler in dry-run mode, not sure if that works, we'll see
2023-01-02 06:38:42 <ikke> fyi, it's on gbr2-dev1
2023-01-02 10:50:59 <ikke> select count(*) from resource_label_events where merge_request_id = 43826 => 19258 😲
2023-01-02 10:53:38 <clandmeter> can you purge them?
2023-01-02 10:53:47 <ikke> I did
2023-01-02 10:54:26 <ikke> I also sent an e-mail
2023-01-02 11:04:10 <mps> clandmeter: linux-edge 6.1.2 for aarch64 now build without CMA
2023-01-02 11:04:43 <clandmeter> mps: thx
2023-01-02 11:05:08 <mps> clandmeter: you are welcome :)
2023-01-03 14:04:39 <minimal> The Jan IRC logs for alpine-linux seem to be missing the correct MIME type (and so browser downloads them rather than displaying them). This happens occasionally at the start of the month when a new logfile is creates
2023-01-03 18:00:44 <clandmeter> minimal: seems only to happen in some specific browsers?
2023-01-03 18:03:17 <clandmeter> on my mac it fails in chrome but works in firefox
2023-01-03 18:03:34 <clandmeter> curl shows the same headers for 2 diff files
2023-01-03 18:10:15 <minimal> clandmeter: not sure but its an infrequent problem which I've seen with the alpine ircs logs in the past
2023-01-03 18:11:18 <minimal> so alpine-devel for this month is fine, alpine-linux however downloads rather than displays
2023-01-03 18:12:00 <minimal> that's on the same browser
2023-01-03 18:40:41 <psykose_> same for me, displays in firefox and downloads on chromium
2023-01-03 18:40:44 <psykose_> content type is text/plain
2023-01-03 18:40:52 <psykose_> i guess chromium should simply stop sucking
2023-01-03 18:41:32 <clandmeter> I will add a text, only support Firefox
2023-01-03 18:41:38 <psykose_> it's the exact same text/plain for working ones
2023-01-03 18:41:39 <clandmeter> :D
2023-01-03 18:42:28 <clandmeter> Is there a beter content type in can use?
2023-01-03 18:45:10 <psykose_> probably not? practically all the various text/* types make firefox download the file instead of displaying lol
2023-01-03 18:47:35 <clandmeter> Maybe it’s wrong encoding?
2023-01-03 18:48:20 <psykose_> they both state the same in headers
2023-01-03 18:48:32 <psykose_> i would be surprised if the actual content is different/lying, but i didn't check
2023-01-03 19:48:21 <clandmeter> could be that chrome analyzes the content and if it differs offers a dl
2023-01-03 20:01:02 <minimal> hmm, perhaps Chrome is indeed trying to be too smart for its own good ;-)
2023-01-03 20:07:02 <pj[m]> still better than not implementing displaying any text/* other than text/plain like firefox
2023-01-04 14:57:06 <mps> <grumble hat on> as alpine becoming more popular quality of bug reports and merge requests goes down </grumble hat off>
2023-01-04 14:57:42 <ikke> patience is a virtue
2023-01-04 14:58:43 <ikke> And empathy
2023-01-04 15:04:32 <mps> hah ;p
2023-01-04 15:05:05 <mps> (so I'm not virtuous)
2023-01-04 18:40:16 <ikke> libtbb test suite us using quite some CPU
2023-01-04 18:41:22 <psykose> it hangs some test like 2% of the time on s390x
2023-01-04 18:41:28 <psykose> and then does that
2023-01-04 18:41:31 <psykose> i guess
2023-01-04 18:42:35 <ikke> 120/121 Test #114: test_malloc_compliance ...................   Passed  211.80 sec
2023-01-04 18:42:41 <ikke> So I suppose hanging on the last test
2023-01-04 18:43:14 <ikke> /home/buildozer/aports/community/libtbb/src/oneTBB-2021.8.0/build/gnu_12.2_cxx11_64_minsizerel/conformance_mutex --force-colors=1
2023-01-04 18:43:18 <psykose> nah, not last
2023-01-04 18:43:54 <ikke> should I kill it?
2023-01-04 18:44:00 <psykose> sure
2023-01-06 07:44:43 <PureTryOut> > error: while writing precompiled header: No space left on device
2023-01-06 07:44:50 <PureTryOut> Uh, is the x86_64 builder out of disk space?
2023-01-06 07:54:57 <ikke> builder or ci runner?
2023-01-06 07:55:07 <ikke> builder is not out of space
2023-01-06 08:51:53 <PureTryOut> sorry CI runner
2023-01-06 12:52:34 <ikke> PureTryOut: made some space on the ci host again
2023-01-06 12:52:49 <PureTryOut> Cool 👍️
2023-01-07 00:19:40 <psykose_> seems all of octave is stuck ikke ^
2023-01-07 04:12:31 <psykose_> ikke: also pressing download on artifacts gives this now https://img.ayaya.dev/QaQv7atTC74D
2023-01-07 06:32:43 <ikke> ouch
2023-01-07 06:32:47 <ikke> 41G log file
2023-01-07 06:33:54 <ikke> 107G even on x86_64 :/
2023-01-07 06:34:18 <ikke> Use of uninitialized value $_ in pattern match (m//) at ./doc/interpreter/mk-qthelp.pl line 30, <$HTML> line 831.
2023-01-07 06:34:50 <psykose_> yeah it's weird
2023-01-07 06:34:56 <psykose_> idk why it looped and filled everything
2023-01-07 06:35:01 <psykose_> i can run them myself fine
2023-01-07 06:59:34 <psykose_> ikke: also riscv is stuck
2023-01-07 07:09:18 <psykose_> also 3.17 has to be restarted cause they got stuck due to diskspace
2023-01-07 07:18:35 <psykose_> also figured out octave issue, was texinfo
2023-01-07 07:18:39 <psykose_> i think
2023-01-07 07:36:03 <psykose_> so yeah just restart everything and it should be fine
2023-01-07 07:36:09 <psykose_> otherwise everything is stuck again
2023-01-07 08:23:13 <psykose> still gotta cycle every active builder
2023-01-07 08:26:03 <ikke> Can do later 
2023-01-07 09:23:17 <psykose> don't forgetti 3.17 too
2023-01-07 09:24:16 <ikke> I don't see anything running there
2023-01-07 09:27:29 <psykose> yeah, it's stuck in git clone
2023-01-07 09:27:42 <psykose> er, pull
2023-01-07 09:27:46 <psykose> from the oom situation
2023-01-07 09:27:52 <psykose> or well, out of disk
2023-01-07 09:27:54 <psykose> words, good
2023-01-07 09:28:22 <psykose> ah, a retry works it seems
2023-01-07 09:28:30 <ikke> not for x86*
2023-01-07 09:29:06 <ikke> oh, now they are
2023-01-07 09:30:24 <psykose> love computers
2023-01-07 09:30:25 <psykose> thanks
2023-01-08 18:21:07 <mps> `make ARCH="arm64" DTC_FLAGS="-@" dtbs` on aarch64 stopped to work. anyone have idea what was changed it tools to cause this
2023-01-08 18:40:01 <mps> hm, with upgrade toolchains it pass but with a lot of warnings about some dts files
2023-01-08 21:14:56 <qontinuum> ikke: thanks for the notification about my server not being reachable (you asked in the end of november so my answer is quiet late but eh). I has some hardware issues that still aren't fixed and it is about my NIC so it can happen that my server goes dark.
2023-01-08 21:15:19 <ikke> ok, thanks for the heads-up
2023-01-08 21:16:01 <ikke> Right now it seems fine
2023-01-08 21:18:20 <qontinuum> Yup, I think it is about my NIC getting overloaded and the driver doing wrong stuff. I keep watching on this anyway because it is really an issue for me since I host some network intensive applications
2023-01-10 05:49:15 <psykose> ikke: do you mind helping me with a wiki migration when you have time
2023-01-10 05:49:35 <psykose> i don't quite remember how to set up some backups first and it's a migration to modern php since they released the version that supports it
2023-01-10 05:59:19 <ikke> Sure, I'll have time later
2023-01-10 06:06:01 <psykose> ah i found the backup command in shell history
2023-01-10 06:08:10 <ikke> :)
2023-01-10 06:09:54 <psykose> also i think the submodule extension checkouts were a little wrong, but i finagled a bit of submodule foreach
2023-01-10 06:10:00 <psykose> hopefully nothing broke :)
2023-01-10 06:10:43 <psykose> added Patroller to go through diffs easier
2023-01-10 06:15:21 <psykose> updated mariadb to 10.6.11 and the container to 3.17
2023-01-10 06:15:38 <psykose> wiki to 1.38.5 for now, will wait for your time to go to .39
2023-01-10 06:16:25 <psykose> how did you put up the frozen db / landing page before?
2023-01-10 06:32:13 <psykose> hm, how does submodule update really work
2023-01-10 06:33:02 <psykose> the .gitmodules only tracks a branch (REL1_38)
2023-01-10 06:33:15 <psykose> but if i do submodule update i get like.. one commit older than the tip of the branch
2023-01-10 06:34:18 <psykose> https://img.ayaya.dev/polFKXmuhsHG
2023-01-10 06:34:22 <psykose> which doesn't really make sense
2023-01-10 11:22:42 <ikke> psykose: the submodule commit is recorded in the parent commit
2023-01-10 11:22:47 <psykose> ah
2023-01-10 11:22:49 <ikke> git submodule update will checkout the commit that is recorded there
2023-01-10 11:22:49 <psykose> ok
2023-01-10 11:23:01 <ikke> That will make sure everyone gets the same commit
2023-01-10 11:30:48 <ikke> So if you want the newer commit, you need to check that out and commit it in the parent repo
2023-01-10 11:33:10 <psykose> yeah. though don't really want to create local commits out of lazyness
2023-01-10 11:33:18 <psykose> that and i guess it's fine to keep it at what they keep it at
2023-01-10 11:33:27 <psykose> maybe something breaks in their submodules at newer :p
2023-01-10 11:34:53 <ikke> yes, certainly
2023-01-10 11:35:48 <psykose> thanks for the explanation!
2023-01-10 11:35:59 <ikke> Np
2023-01-10 15:55:06 <ncopa> ikke: any idea why this fails in CI? https://gitlab.alpinelinux.org/alpine/infra/docker/abuild-ci/-/merge_requests/2
2023-01-10 16:08:17 <clandmeter> ncopa: looks like the runners are not abled?
2023-01-10 16:08:22 <clandmeter> enabled*
2023-01-10 16:08:47 <clandmeter> https://gitlab.alpinelinux.org/nmeum/abuild-ci/-/settings/ci_cd
2023-01-10 16:10:59 <nmeum> do I need to enable it explicitly in my fork?
2023-01-10 16:11:43 <clandmeter> i think so
2023-01-10 16:12:37 <nmeum> shared runners are enabled in the fork
2023-01-10 16:12:47 <clandmeter> yes but not the docker ones
2023-01-10 16:12:58 <clandmeter> they are special I believe
2023-01-10 16:13:17 <clandmeter> you can however just commit to a featuer branch on the project itself
2023-01-10 16:14:25 <clandmeter> looks like they are also not enabled on the project itself
2023-01-10 16:17:10 <clandmeter> if you open the failed job, it is part of your own namespace, and it shows that it needs Tags: docker-alpine ci-docker-image aarch64
2023-01-10 16:17:42 <clandmeter> i see you already enabled them :)
2023-01-10 16:23:03 <ikke> gitlab security release
2023-01-10 16:23:54 <psykose> fun
2023-01-10 16:24:10 <ikke> Seems to be mild, highest severity is medium
2023-01-10 16:24:15 <ikke> https://about.gitlab.com/releases/2023/01/09/security-release-gitlab-15-7-2-released/
2023-01-10 16:28:42 <psykose> the only ones even remotely relevant is the regex one and the group revoke one
2023-01-10 19:06:03 <ikke> nmeum: I created a new pipeline, they do run now
2023-01-10 19:06:21 <ikke> ncopa: ^
2023-01-10 20:16:41 <psykose> ikke: i noticed on new gitlab that merging anything always kills the pipeline it seems :)
2023-01-10 20:16:44 <psykose> is nice
2023-01-10 20:19:32 <ikke> yes
2023-01-10 20:20:14 <ikke> although, aports-qa-bot also does something regarding that
2023-01-10 20:20:21 <ikke> so not sure if it's gitlab or aports-qa-bot
2023-01-10 20:20:23 <psykose> :)
2023-01-10 20:20:32 <psykose> and glab merge got even worse
2023-01-10 20:20:37 <ikke> oh, great
2023-01-10 20:20:43 <psykose> fails sometimes even on no rebase half the time
2023-01-10 20:20:50 <psykose> still wonder what's up with that
2023-01-10 20:21:11 <psykose> i miss 2 versions ago when it would just always succeed
2023-01-10 20:22:18 <ikke> We should open up an issue on that
2023-01-10 20:22:37 <psykose> hmm, yeah
2023-01-10 20:22:42 <psykose> you can reproduce it right
2023-01-10 20:23:06 <psykose> (just instantly returns on glab mr merge --rebase --yes --when-pipeline-succeeds=false)
2023-01-10 20:23:07 <ikke> yes, happened for me as well
2023-01-10 20:23:17 <ikke> though been a while since I used it
2023-01-10 20:23:27 <ikke> new gitlab also has rebase without pipeline
2023-01-10 20:23:44 <psykose> yeah, and glab doesn't expose it
2023-01-10 20:23:56 <ikke> right, so I use the webif more lately
2023-01-10 20:24:03 <psykose> i mostly got used to the tool because it's one action instead of two
2023-01-10 20:24:19 <psykose> (paste mr number, done, instead of rebase, refresh page a bunch times, merge)
2023-01-10 20:24:36 <ikke> nod
2023-01-10 20:24:52 <psykose> but nowadays it's more like 5 actions
2023-01-10 20:25:06 <psykose> and i even had a merge; sleep 30; merge in tehre
2023-01-10 20:25:08 <psykose> :p
2023-01-10 20:25:34 <psykose> (i have it macrod to `gmm $id`)
2023-01-11 01:44:37 <tomalok> ikke - looks like https://alpinelinux.org/releases.json has an incorrect date for 3.17.1  (lists 2022-01-09 instead of 2023...)  was wondering why the cloud image builder wasn't finding it (it reverse sorts by date and picks the most recent).
2023-01-11 05:57:03 <ikke> tomalok: apparently ncopa fixed the date in the post, but releases.json still lists the old date
2023-01-11 06:11:59 <ikke> tomalok: should be fixed now
2023-01-11 11:57:15 <ikke> maintenance anounced for deu1 and dmvpn1
2023-01-11 11:58:08 <ikke> wait, for 2020-08-08..
2023-01-11 11:59:18 <ikke> blaboon: 2 of our linodes show maintenance scheduled for maintenances that have been completed already
2023-01-11 12:01:31 <psykose> it's 2020-08-07
2023-01-11 12:01:41 <psykose> ;)
2023-01-11 12:03:10 <ikke> Oh, in that case they will be maintenanced tomorrow
2023-01-11 14:51:29 <blaboon> ikke: i took a look at the linodes in question and the physical hosts they're on and couldn't find any upcoming scheduled maintenance for either one, so it sounds like maybe something is buggy somewhere (perhaps an API issue). i would open up a ticket to our support team so that they can escalate to the appropriate team
2023-01-11 14:52:08 <ikke> 👍
2023-01-11 15:00:41 <blaboon> it looks like an old maintenance window from several years ago might have been re-activated
2023-01-11 15:05:01 <blaboon> it seems like our support team is already aware of the issue, several other customers have already opened tickets for the same thing
2023-01-11 15:05:46 <ikke> Alright
2023-01-11 15:06:59 <blaboon> seems like the root cause of the issue is that the API now returns both current and past events when you query the maintenance endpoint, and the cloud manager hasn't been updated to filter out events that have already passed. the dev team is working on getting that fixed
2023-01-11 15:11:59 <PureTryOut> Oh hey build.a.o shows the amounts of things it's building now, nice!
2023-01-11 15:13:25 <ikke> Yes
2023-01-11 15:13:52 <ikke> Courtesy of ptrc/psykose 
2023-01-11 15:24:43 <PureTryOut> Thanks for that!
2023-01-11 15:25:46 <ptrc> i'm glad someone else also finds it useful :)
2023-01-11 20:41:32 <pj[m]> actually when I need you, you decided to die
2023-01-11 20:41:37 <pj[m]> smh
2023-01-11 20:41:43 <ikke> git.a.o?
2023-01-11 20:41:49 <ikke> it was just briefly unreahcable apparently
2023-01-11 20:41:57 <ikke> When I checked it was reachable
2023-01-11 20:42:06 <pj[m]> it dies for me
2023-01-11 20:42:16 <pj[m]> 504
2023-01-11 20:42:21 <ikke> Hmm, ok, now it's gone for me as wel
2023-01-11 20:44:44 <ikke> restarted it
2023-01-11 20:51:06 <pj[m]> thank
2023-01-11 22:09:50 <pj[m]> seems still kinda dead
2023-01-11 22:10:12 <ikke> load seems pretty high on that machine
2023-01-11 22:15:15 <ikke> pj[m]: should improve now
2023-01-11 22:15:46 <pj[m]> thank again
2023-01-11 22:16:02 <pj[m]> wouldn't mind having gitlab link on pkgs.a.o
2023-01-11 22:16:20 <ikke> yeah, was thinking about that as well, but never bothered enough to actual change it
2023-01-11 22:17:19 <pj[m]> thought it was already implemented by ptrc
2023-01-11 22:19:13 <ikke> Mostly changed things regarding anitya
2023-01-11 22:19:26 <ikke> and changing bugs.a.o to gitlab.a.o
2023-01-11 22:20:43 <ikke> I'll add a todo to look at it when I have time
2023-01-12 03:10:58 <psykose> someone already complained that some link somewhere (alpinelinux.org top right git?) pointed to gitlab instead of cgit
2023-01-12 03:11:07 <psykose> can't imagine what they would say if they all pointed to gitlab
2023-01-12 06:31:26 <ikke> Heh
2023-01-12 11:10:49 <ncopa> I'd like to rename dl-cdn.alpinelinux.org/alpine/ -> cdn.alpinelinux.org/ Is that realistic?
2023-01-12 11:11:12 <ncopa> so we have https://cdn.alpinelinux.org/v3.17/main
2023-01-12 11:11:27 <ncopa> actually
2023-01-12 11:11:45 <ncopa> i'd like it to be something like: https://cdn.alpinelinux.org/3.17-stable/main
2023-01-12 11:11:59 <ncopa> where 3.17-stable is equal to the git branch
2023-01-12 11:12:08 <ikke> We probably need to keep dl-cdn as a cname for quite some time
2023-01-12 11:12:12 <ncopa> yes
2023-01-12 11:12:15 <ptrc> the old hostname would have to stay there for at least a few years
2023-01-12 11:12:18 <ncopa> yes
2023-01-12 11:12:28 <ikke> Make sure cdn.a.o is added to the fastly certificate
2023-01-12 11:12:29 <ncopa> i'd say 2-3 year
2023-01-12 11:12:35 <ikke> I'd say longer
2023-01-12 11:12:42 <ikke> docker images can stay around quite a long time
2023-01-12 11:12:55 <ncopa> 4-5 years?
2023-01-12 11:12:55 <ikke> servers with old installations
2023-01-12 11:13:04 <ncopa> true
2023-01-12 11:13:10 <ncopa> maybe we can never remove the old hostnames
2023-01-12 11:13:26 <ikke> We always kept most dns records
2023-01-12 11:13:32 <ikke> and pointed them to somewhere
2023-01-12 11:13:42 <ptrc> i've seen people still using nl.a.o and iirc it was removed from the list quite some time ago
2023-01-12 11:13:47 <ikke> yes
2023-01-12 11:14:14 <ncopa> but what do you think, is cdn.a.o a good or bad idea?
2023-01-12 11:14:28 <ncopa> and getting rid of the /alpine/ subdirectory
2023-01-12 11:14:40 <ncopa> and rename v3.17 -> 3.17-stable
2023-01-12 11:14:55 <ncopa> and maybe also rename git branch `master` to `edge`
2023-01-12 11:15:04 <ikke> as long as wel have symlinks as fallback
2023-01-12 11:15:20 <ncopa> symlinks as fallbacks is a must ofc
2023-01-12 11:15:24 <ptrc> git branch rename doesn't make much sense imo
2023-01-12 11:15:32 <ncopa> i think the most problematic thing is git branch rename
2023-01-12 11:15:52 <ncopa> the idea is to have git branch name = release branch name
2023-01-12 11:16:09 <ncopa> instead of master -> edge and 3.17-stable -> v3.17
2023-01-12 11:16:43 <psykose> ptrc: nl.a was removed like last year or a bit more
2023-01-12 11:18:14 <psykose> ncopa: it doesn't sound bad but as a constructionist of bike sheds what is the real gain? we mostly get a 1:1 mapping but there is nothing else except short- and medium-term confusion for people
2023-01-12 11:19:22 <psykose> i.e. especially in git i don't think it's going to be possible unless git has a way to symlink it itself
2023-01-12 11:19:31 <psykose> there's probably a million places referencing master out there
2023-01-12 11:19:56 <psykose> it would be easier to rename edge to master :p
2023-01-12 11:20:25 <ptrc> psykose: ln -s master .git/refs/heads/edge
2023-01-12 11:20:30 <ptrc>  /s
2023-01-12 11:20:45 <ikke> git does have symbolic refs :)
2023-01-12 11:21:25 <ikke> fun
2023-01-12 11:21:48 <ncopa> the idea is to clarify the edge -> master and v3.17 -> 3.17-stable relation
2023-01-12 11:22:03 <ncopa> makes it very simple that release branch = git branch
2023-01-12 11:22:13 <ncopa> it also lets us simplify code some places
2023-01-12 11:22:22 <ikke> cgit is using 100% cpu
2023-01-12 11:23:12 <ncopa> we could rename edge -> master, but the term "master" is a bit controversial, and I would like to avoid the drama that potentially would follow
2023-01-12 11:23:27 <ncopa> because "alpine edge" is a well known concept at this point
2023-01-12 11:23:35 <ikke> I don't think it would make sense anyway
2023-01-12 11:24:04 <ncopa> I also find the term "edge" more clean than "master"
2023-01-12 11:24:13 <ncopa> edge, 3.17-stable, 3.16-stable
2023-01-12 11:24:19 <ncopa> its crystal clear to me
2023-01-12 11:24:47 <ikke> as long as we don't break existing setups, few people would care
2023-01-12 11:25:02 <ncopa> otoh, the term "edge" is used in "edge" computing
2023-01-12 11:25:06 <ncopa> just tinking out loud here now
2023-01-12 11:25:23 <ncopa> maybe something like "rolling" makes more sense?
2023-01-12 11:25:29 <ncopa> or "head" or "dev"
2023-01-12 11:25:44 <ikke> unstable
2023-01-12 11:25:56 <ncopa> yeah, thats also a good candidate
2023-01-12 11:26:20 <ikke> but I'd still prefer edge
2023-01-12 11:26:30 <ncopa> yeah, i like "edge"
2023-01-12 11:26:53 <ncopa> and yes, we will not do it unless we can do it without braking stuff
2023-01-12 11:27:04 <ncopa> breaking backwards compat
2023-01-12 11:29:12 <mps> 3.17-stable? isn't 3.17 already stable
2023-01-12 11:29:35 <ncopa> it is
2023-01-12 11:30:15 <ncopa> but it would be more visible
2023-01-12 11:30:44 <ikke> mps: the goal is to match the branch-name
2023-01-12 11:30:54 <mps> but we don't have 3.17-unstable, or edge or anything
2023-01-12 11:31:04 <ikke> mps: we have a branch called 3.17-stable
2023-01-12 11:31:16 <mps> yes, that true
2023-01-12 11:31:20 <ncopa> we currently have branch 'master' and '3.17-stable' ...
2023-01-12 11:31:32 <mps> yes, I know
2023-01-12 11:31:54 <ncopa> so the other option is to rename 3.17-stable branch to v3.17
2023-01-12 11:31:55 <mps> but not 3.17-stable release
2023-01-12 11:32:32 <ikke> v3.17 (specifically starting with a v) can be confusing with tags
2023-01-12 11:32:33 <ncopa> ncopa> the idea is to clarify the edge -> master and v3.17 -> 3.17-stable relation
2023-01-12 11:32:33 <ncopa> ncopa> makes it very simple that release branch = git branch
2023-01-12 11:33:10 <ncopa> ikke i agree, which is why i think 'edge', '3.17-stable', '3.n-stable' .... makes most sense
2023-01-12 11:33:23 <mps> just contemplating about naming
2023-01-12 11:33:38 <mps> I'm not against change to be clear
2023-01-12 11:39:11 <ncopa> so we could use `git symbolic-ref` to rename the master branch? https://git-scm.com/docs/git-symbolic-ref
2023-01-12 11:39:24 <mps> hope we will not have code names for releases, like woody, potato, stretch .... ;-)
2023-01-12 11:39:45 <ncopa> mps: agree
2023-01-12 11:40:44 <mps> 
ACTION used debian for more than 20 years and never wasn't sure which numerical version I use

2023-01-12 11:44:37 <psykose> ikke: yeah sometimes it runs into specific queries that take forever
2023-01-12 11:44:54 <psykose> originally it was just 1 thread and blocked on queries, so 4 was really just masking the issue
2023-01-12 11:45:11 <psykose> or well 3
2023-01-12 11:46:27 <psykose> and the container is 3.14 too since newer versions break the git clone and always 429
2023-01-12 12:02:36 <mps> anyone have idea why commit cdd3ce01ff79a74ae2e87c50ecdc5bbb358d4df6 touched linux-edge
2023-01-12 12:35:51 <ikke> "collapse newlines "
2023-01-12 12:38:16 <ikke> nothing significant or impacting
2023-01-12 12:38:53 <mps> I added warning note at the top of APKBUILD to resist touching linux-edge without urge
2023-01-12 12:39:24 <ikke> Not that it would matter with these kinds of bulk updates
2023-01-12 12:40:23 <mps> well, it requires more work when cherry pick upgrade from edge to latest stable
2023-01-12 12:40:39 <mps> and this could make mess
2023-01-12 12:47:11 <ikke> I mean is that these kinds of bulk-updates are automated, no one is reading any comments of individual APKBUILDs
2023-01-12 12:59:40 <mps> ikke: I tend to think same as you
2023-01-12 13:00:12 <mps> because this I don't like bulk updates
2023-01-12 13:04:46 <pj[m]> "head" branch would be confusing with git HEAD
2023-01-12 13:06:01 <ikke> yes
2023-01-12 13:16:59 <psykose> ikke: could you help me figure out why cloning https://git.alpinelinux.org/aports 429's most of the time
2023-01-12 13:17:04 <psykose> i am not sure what layer this is at
2023-01-12 13:18:07 <psykose> (it's on any version newer than 3.14 of alpine somehow)
2023-01-12 13:19:59 <ikke> hmm
2023-01-12 13:20:17 <ikke> Dues uwsgi has some auto circuit breaker that returns 429?
2023-01-12 13:20:43 <psykose> maybe
2023-01-12 13:20:43 <psykose> hmm
2023-01-12 13:20:58 <psykose> but .14 and .15 have the same version
2023-01-12 13:21:07 <psykose> and it broke before just between those
2023-01-12 13:21:37 <psykose> also doesn't seem like it does
2023-01-12 13:21:42 <psykose> nginx config doesn't do anything either
2023-01-12 13:22:23 <ikke> only for one specific user agent
2023-01-12 13:24:42 <psykose> yeah
2023-01-12 13:25:05 <psykose> does traefik do anything fancy based on heuristic?
2023-01-12 13:27:23 <psykose> maybe it's git-daemon internals
2023-01-12 13:27:27 <ikke> Not that I'm aware of
2023-01-12 13:27:34 <ikke> but I don't know the intrinsics of traefik that well
2023-01-12 13:27:53 <ikke> If you enable json output in traefik, it does give more details
2023-01-12 13:28:29 <psykose> is there an easy way to get traefik logs for only cgit
2023-01-12 13:28:40 <psykose> git-daemon default --max-connections is 32
2023-01-12 13:30:22 <ikke> don't think so
2023-01-12 13:34:50 <psykose> doesn't seem to log anything even with loglevel info
2023-01-12 13:39:23 <ikke> did you set format to json?
2023-01-12 13:39:49 <ikke> https://tpaste.us/x5PW 
2023-01-12 13:45:23 <psykose> just log not accesslog
2023-01-12 13:45:24 <psykose> thanks
2023-01-12 13:46:51 <psykose> idk why it would log more with that
2023-01-12 13:47:30 <psykose> doesn't say "we log nothing unless json", just says a different format is default
2023-01-12 13:47:38 <psykose> also this logs like a thousand things per second holy
2023-01-12 13:48:22 <psykose> "DownstreamStatus":429
2023-01-12 13:48:26 <psykose> ok, so it comes from the service
2023-01-12 13:48:46 <psykose> people sure love pulling the secdb it seems
2023-01-12 13:49:12 <psykose> so, it's either git-daemon or it's nginx
2023-01-12 13:50:34 <psykose> nginx logs say it's 200 for the response
2023-01-12 13:50:36 <psykose> ????
2023-01-12 13:59:44 <ikke> yeah, secdb is popular, 40 requests per second
2023-01-12 14:00:43 <clandmeter> we should create plans for subscription
2023-01-12 14:01:22 <psykose> hmm what if
2023-01-12 14:01:42 <psykose> traefik.http.middlewares.cgit-ratelimit.ratelimit.burst=2 is too low and a clone needs like 3 for no reason suddenly
2023-01-12 14:01:51 <psykose> also i don't think this git-daemon thing is actually used at all
2023-01-12 14:02:22 <psykose> i made it 4 and it always works now
2023-01-12 14:02:23 <psykose> epic
2023-01-12 14:02:54 <psykose> well at least i wrote out a custom build of cgit for the uwsgi container
2023-01-12 14:02:56 <psykose> something productive
2023-01-12 14:03:03 <psykose> ikke: got a moment for mediawiki?
2023-01-12 14:36:45 <ikke> psykose: sure
2023-01-12 14:38:25 <psykose> oki, freeze it and landing page it
2023-01-12 14:38:39 <ikke> How can I help?
2023-01-12 14:38:52 <psykose> (since i don't know how you normally do that)
2023-01-12 14:39:02 <ikke> most of the time, we don't
2023-01-12 14:39:08 <ikke> depends on how long the down time
2023-01-12 14:39:15 <psykose> probably a bit
2023-01-12 14:39:18 <psykose> php version and all
2023-01-12 14:39:22 <psykose> something might explode :p
2023-01-12 14:39:23 <ikke> Do you have a nice landing page?
2023-01-12 14:39:28 <psykose> nope
2023-01-12 14:39:31 <psykose> idk, you did something last time
2023-01-12 14:39:38 <ikke> maintenace I gues
2023-01-12 14:39:40 <psykose> yea
2023-01-12 14:40:26 <ikke> anoying, someone set the default ssh config to disallow tunneling / jumping
2023-01-12 14:40:49 <ikke> AllowTcpForwarding no
2023-01-12 14:40:57 <psykose> hmm
2023-01-12 14:41:23 <psykose> might have been me by accident on update-conf but not sure
2023-01-12 14:41:33 <ikke> I have seen it on multiple servers
2023-01-12 14:41:34 <psykose> i think it was there for a while
2023-01-12 14:41:36 <psykose> ah
2023-01-12 14:41:46 <psykose> it is the default alpine conf after all
2023-01-12 14:41:50 <ikke> it was not
2023-01-12 14:42:29 <psykose> 495bbd7fb1f07c23a1f2d47a071aa5519e08744c
2023-01-12 14:42:45 <ikke> hmm
2023-01-12 14:42:48 <ikke> a while ago
2023-01-12 14:42:51 <psykose> ye
2023-01-12 14:43:33 <ikke> At the top of LocalSettings.php $wgReadOnly
2023-01-12 14:47:35 <psykose> doesn't seem to change the landing :p though i think it works
2023-01-12 14:47:44 <ikke> It makes it read-only
2023-01-12 14:48:03 <ikke> A read landing page is probably something custom
2023-01-12 14:56:49 <psykose> the.. update fails if  you have readonly set
2023-01-12 14:56:50 <psykose> haha
2023-01-12 14:57:00 <ikke> funny
2023-01-12 14:57:13 <psykose> ok i think it worked
2023-01-12 14:57:21 <psykose> everything is running
2023-01-12 14:58:10 <psykose> also the logs are constantly spammed with the tty5/6 gettys dying in a loop
2023-01-12 14:58:33 <ikke> I suppose we need to change something in /etc/inittab
2023-01-12 14:59:23 <psykose> it doesn't have 5/6 in it
2023-01-12 14:59:24 <psykose> hm
2023-01-12 15:01:37 <psykose> anyway seems to run fine
2023-01-12 15:01:43 <psykose> has the error thumbnail though
2023-01-12 15:04:26 <tomalok> ikke: thanks for updating releases.json -- i've built/published the 3.17.1 cloud images -- https://gitlab.alpinelinux.org/alpine/infra/alpine-mksite/-/merge_requests/59 updates the website
2023-01-12 15:09:37 <psykose> i forget what fixed the thumbnails before..
2023-01-12 15:12:31 <ikke> disabling readonly mode
2023-01-12 15:14:57 <psykose> it is
2023-01-12 15:15:24 <psykose> what
2023-01-12 15:15:25 <ikke> ctrl+f5
2023-01-12 15:15:29 <psykose> ....
2023-01-12 15:15:33 <psykose> it was off this whole time
2023-01-12 15:15:37 <ikke> lol
2023-01-12 15:15:37 <psykose> but enabling it again
2023-01-12 15:15:42 <psykose> and turning it off again..
2023-01-12 15:16:03 <psykose> i guess if you start the update with it on it breaks something
2023-01-12 15:16:21 <psykose> everything else just worked
2023-01-12 15:16:28 <psykose> ah well
2023-01-12 15:16:32 <psykose> all done
2023-01-12 15:16:33 <psykose> thanks
2023-01-12 15:16:37 <psykose> i set up weekly db backups too
2023-01-12 15:16:43 <psykose> /etc/periodic/weekly
2023-01-12 15:17:10 <ikke> ok
2023-01-12 15:18:50 <nmeum> could I get access to an armv7 container? I would like to debug a segfault which only occurs on arm and doing so in qemu-user is always a bit cumbersome since you have to use qemu's gdb stub then
2023-01-12 15:18:53 <nmeum> I have wireguard vpn access
2023-01-12 15:19:29 <ikke> sure, hold on
2023-01-12 15:21:09 <nmeum> 🙏
2023-01-12 15:27:09 <ikke> nmeum: should I add a key from here? https://gitlab.alpinelinux.org/nmeum.keys
2023-01-12 15:27:47 <nmeum> yes, preferably the second one
2023-01-12 15:28:52 <ikke> can you try nmeum-edge-armv7.usa9.alpin.pw (172.16.23.125)
2023-01-12 15:29:35 <psykose> love the <edkey> <edkey> <GIGANTIC WALL OF TEXT> on every list with rsa in it :D
2023-01-12 15:30:08 <ikke> afk for a bit
2023-01-12 15:31:09 <nmeum> ikke: works. thanks a lot! (:
2023-01-12 16:02:21 <mcrute> would it be possible to get a shell account on an aarch64 system? trying to debug an aarch64 only failure but have no hardware to test with at the moment
2023-01-12 16:02:37 <mcrute> aport build failure, that is
2023-01-12 16:08:01 <tomalok> mcrute: i typically do my aports work on EC2 instances, i hear alpine has cloud images for aarch64 instance types these days... ;)
2023-01-12 16:08:33 <mcrute> tomalok: cloud? bah!
2023-01-12 16:09:07 <tomalok> heh
2023-01-12 16:09:28 <minimal> I do my various arch package builds using qemu-user ;-)
2023-01-12 16:09:52 <minimal> together with binfmt
2023-01-12 16:10:03 <mcrute> the funny thing is I actually do have a few aarch64 servers sitting in the datacenter right now they're just in a new build-out colo at the moment so not accessible :-(
2023-01-12 16:10:57 <mcrute> IME emulating aarch64 on Intel with qemu-user is just way to slow for large builds
2023-01-12 16:11:07 <psykose> yeah it's painful
2023-01-12 16:11:11 <psykose> curious, what fails
2023-01-12 16:12:14 <mcrute> a node memory segfault when packing javascript assets in the vault package, no stack trace or anything just throws a std::bad_alloc (IIRC) and fails
2023-01-12 16:13:03 <mcrute> psykose: here's the build log https://build.alpinelinux.org/buildlogs/build-edge-aarch64/community/vault/vault-1.12.2-r0.log
2023-01-12 16:13:10 <psykose> ah that
2023-01-12 16:13:19 <psykose> mm
2023-01-12 16:13:30 <mcrute> I ended up just disabling UI for aarch64, which is fine because the core functionality works, but I would really like to re-enable it
2023-01-12 16:13:38 <mcrute> you know this error?
2023-01-12 16:13:45 <psykose> not the solution to it
2023-01-12 16:13:51 <mcrute> :-(
2023-01-12 16:13:52 <psykose> but it's very common for ember building anything
2023-01-12 16:14:24 <psykose> these large nodejs builds sometimes need playing with --max_old_space_size but i'm not entirely sure how it works
2023-01-12 16:14:39 <psykose> i think when i raised it to 4096 for that it just segfaulted instead, and anything else fails with different errors
2023-01-12 16:14:47 <psykose> (yes i already tried to debug that after seeing the failure)
2023-01-12 16:15:09 <mcrute> I saw similar suggestions in GitHub issues, I just need some aarch64 hardware to poke at it a bit, don't want to burn builder time with it
2023-01-12 16:15:11 <minimal> for the Nomad package I don't build the UI from scratch, just use the provided assets instead
2023-01-12 16:15:22 <minimal> can;t you do the same for Vault?
2023-01-12 16:15:34 <psykose> but the behaviour is strange enough it makes me think there is some weird failure on the aarch64 nodejs
2023-01-12 16:15:38 <mcrute> minimal: oh... possibly
2023-01-12 16:15:41 <ikke> grafana does the same
2023-01-12 16:15:41 <psykose> and it might end up being an actual node bug on musl
2023-01-12 16:15:56 <psykose> or just aarch64 in general
2023-01-12 16:16:02 <ikke> mcrute: Would it be possible for your to setup a wg tunnel?
2023-01-12 16:16:06 <psykose> don't think it makes sense for js to throw a c++ error :p
2023-01-12 16:16:10 <minimal> mcrute: the fundamental problem is that Hashicorp are slow to support newer Node versions
2023-01-12 16:16:13 <mcrute> ikke: yeah, very easily in fact
2023-01-12 16:16:45 <mcrute> minimal: yeah I noticed, went down a path of trying to do the upgrade and it looked like a long road to make that work so I abandoned it
2023-01-12 16:16:49 <ikke> mcrute: https://gitlab.alpinelinux.org/alpine/infra/infra/-/wikis/Alpine-wireguard-VPN
2023-01-12 16:16:51 <minimal> so there was a problem a while back that the new Node LTS came out but Hashicorp didn't support it at that time
2023-01-12 16:17:16 <minimal> so using the assets avoids that whole issue
2023-01-12 16:18:17 <ikke> mcrute: You can use 172.16.252.15/24 as address
2023-01-13 06:19:50 <psykose> :)
2023-01-13 06:20:15 <psykose> algitbot: ping
2023-01-13 15:48:08 <mcrute> ikke: does the aarch64 builder host have 160 cores like the one I have shell access to?
2023-01-13 15:49:19 <ikke> It's the same host
2023-01-13 15:49:50 <mcrute> ah okay, is the core-count situation similar for x86 and armhf?
2023-01-13 15:50:00 <ikke> We do limit containers to 80 cores though (Numa domain) 
2023-01-13 15:50:07 <mcrute> I'm pretty sure now that the node failures I was seeing for vault were because it was creating a thread per visible core and trying to saturate the machine
2023-01-13 15:50:13 <ikke> x86 has less cores 
2023-01-13 15:50:26 <ikke> armhf is same as aarch64 
2023-01-13 15:50:59 <mcrute> well that might explain this then... if I clamp job count to 4 builds are slower but succeed... I guess I'll push this hack to all the builders and see if it works
2023-01-13 15:53:09 <ikke> What is the build failure? 
2023-01-13 15:53:35 <ikke> And it happens from time to time we need to clamp the amount of jobs 
2023-01-13 15:53:42 <mcrute> std::bad_alloc then crash when building assets using ember
2023-01-13 15:54:09 <ikke> OK, so probably using too much memory
2023-01-13 15:55:24 <mcrute> yeah but limiting memory with node flags still results in crashes, also machine memory use doesn't spike or anything so I guess there are other limits
2023-01-13 15:55:47 <mcrute> but when I force the thread-count to 4 vs the 160 it tries to run builds succeed
2023-01-13 15:56:48 <mcrute> do we normally apply the limits in the APKBUILD or somewhere externally?
2023-01-13 15:58:34 <ikke> In /etc/abuild.conf $JOBS is exported
2023-01-13 15:59:19 <ikke> If you want to limit that, you'd do it in the APKBUILD yes
2023-01-13 15:59:48 <mcrute> okay, yeah looks like openjdk and a few packages in testing are doing that... I'll go with that approach
2023-01-14 03:21:49 <psykose> not really sure what is up with that but disabling cgit blame now has everything back to 0 consistently :p
2023-01-14 03:22:13 <psykose> quite literally almost every request is a git blame on a random file, definitely legit for sure
2023-01-14 06:21:55 <psykose> lol
2023-01-14 07:52:30 <psykose> should still probably fix that date format
2023-01-14 07:52:32 <psykose> forgot where it was
2023-01-14 07:52:39 <psykose> also the error was a stuck git lock
2023-01-14 07:52:57 <psykose> afaict the volume is not shared with anything so not sure how that happened
2023-01-14 07:53:16 <psykose> maybe the last restart was timed with the lock and it didn't drop it
2023-01-14 07:53:18 <psykose> fun stuff
2023-01-14 07:53:29 <psykose> even containers need clean exits
2023-01-14 07:54:09 <psykose> that hook also prints like a million errors when it runs about branches not existing :D
2023-01-14 07:55:10 <psykose> ah, the fetch has no --prune
2023-01-14 07:55:11 <psykose> added
2023-01-14 09:53:09 <ikke> psykose: thank you for taking care of cgit 
2023-01-14 09:53:48 <psykose> i assume it's really just someone's dos because it doesn't make sense to git blame "random aport in 3.11-stable" at random every second
2023-01-14 09:54:11 <psykose> i checked the docs but could not find a way to make it show "use gitlab for blame" sadly
2023-01-14 09:55:00 <psykose> aside from that not sure what else i could improve
2023-01-14 09:55:05 <ikke> Not sure if we want to put that load on gitlab to be honest 
2023-01-14 09:55:16 <ikke> Blame is expensive 
2023-01-14 09:55:20 <psykose> yes
2023-01-14 09:55:24 <psykose> especially in a repo like aports
2023-01-14 09:55:29 <ikke> Nod
2023-01-14 09:55:38 <psykose> at best i think it will be a bit better cached if anything (it's not in cgit), but
2023-01-14 09:56:09 <psykose> eh, guess we'll see if someone complains
2023-01-14 09:56:14 <psykose> i've never used the cgit blame before
2023-01-14 09:56:22 <psykose> it's more of a developer feature to begin with
2023-01-14 09:56:38 <psykose> and developers probably have a checkout..
2023-01-14 09:56:51 <ikke> Hmm, maybe we should make sure commit graphs are enabled 
2023-01-14 09:57:15 <psykose> they are
2023-01-14 09:57:35 <psykose> not sure how they are shown somewhere though
2023-01-14 09:58:02 <ikke> No, it's just a feature uses in the back to speed up certain operations 
2023-01-14 09:58:26 <psykose> ah
2023-01-14 09:58:35 <psykose> enable-commit-graph
2023-01-14 09:58:35 <psykose> Flag which, when set to "1", will make cgit print an ASCII-art commit history graph to the left of the commit messages in the repository log page. Default value: "0".
2023-01-14 09:58:54 <psykose> maybe not this
2023-01-14 10:00:31 <ikke> No, that would be more like git log --graph
2023-01-14 10:00:45 <ikke> It's a config setting in the repo
2023-01-14 10:01:01 <ikke> And git gc / maintenance run is what generates it 
2023-01-14 10:01:12 <psykose> ah, gitrc?
2023-01-14 10:01:25 <ikke> .git/config
2023-01-14 10:01:30 <psykose> yeah
2023-01-14 10:01:33 <psykose> there's enable-git-config to read that too
2023-01-14 10:03:02 <psykose> gc.writeCommitGraph is default true
2023-01-14 10:03:07 <psykose> i think all of that is default
2023-01-14 10:03:18 <psykose> fetch.writeCommitGraph is not though
2023-01-14 10:04:13 <psykose> Having an updated commit-graph file helps performance of many Git commands, includin ggit merge-base, git push -f, and git log --graph. Defaults to false.
2023-01-14 10:04:43 <psykose> i guess i can add that to the rc but i'm not sure if it will help for the specific view-file usecases
2023-01-14 10:05:24 <psykose> i will disable the cgit graph though as that's not really useful at all for something like aports
2023-01-14 10:05:59 <psykose> (it's the thing on the left that can show where other branches are, which is.. never the case given how far they get from each other lol)
2023-01-14 10:06:18 <psykose> less visual clutter too
2023-01-14 10:06:29 <psykose> how should i maintain the .git/config in this?
2023-01-14 10:06:44 <psykose> also you have to kick/restart edge-armv7 because it's stuck
2023-01-14 10:09:39 <ikke> psykose: forgot where it is stored, but perhaps a bind mount? 
2023-01-14 10:09:58 <psykose> it's in a git: volume in compose
2023-01-14 10:10:19 <psykose> but i'm just wondering if there's something better than "i enter the container and git config the options once"
2023-01-14 10:10:51 <ikke> Bind mounts 
2023-01-14 10:11:01 <ikke> That's how we typically do that 
2023-01-14 10:11:05 <psykose> oh you mean like
2023-01-14 10:11:07 <psykose> i put it in a folder
2023-01-14 10:11:11 <psykose> then move it into the thing
2023-01-14 10:11:12 <psykose> hm
2023-01-14 10:11:34 <psykose> ok, so global git config mounted
2023-01-14 10:12:24 <psykose> can't tell if this runs as root or git
2023-01-14 10:15:51 <psykose> ah, /etc/gitconfig is a thing
2023-01-14 10:16:44 <psykose> ok, added fetch writecommitgraph, anything else you think is good option-wise
2023-01-14 10:22:32 <psykose> technically the hourly gc auto which does run on fetch afaik has commit-graph already but
2023-01-14 11:40:26 <ikke> There is also maintenance run, which does more then just gc afaik
2023-01-14 11:40:29 <ikke> not sure if it's helpful
2023-01-14 11:48:46 <psykose> by default e.g. something like fetch (manpage doesn't specify full set of commands) all trigger maintenance run ..something
2023-01-14 11:48:51 <psykose> and there are timers for everything
2023-01-14 11:49:09 <psykose> so, the hourly maintenance does commit-graph, weekly there is a pack-ref ..
2023-01-14 11:49:23 <psykose> it only runs on command invocation not as a daemon but there are constant fetches anyway
2023-01-14 11:49:48 <ikke> right
2023-01-14 11:49:58 <psykose> perhaps there are some that never run on any timeframe that are useful, but there are quite a few i think
2023-01-14 11:51:05 <ikke> scheduling it mostly makes sure it does not need to run just when you need to do something
2023-01-14 11:51:59 <psykose> ah, maintenance.gc.enabled is the only true one.. touching some stuff sets the others
2023-01-14 11:52:32 <psykose> doing register once will fill it out
2023-01-14 11:53:15 <psykose> ok, i'll copy some config for it, but passive maintenance requires cron
2023-01-14 11:53:31 <psykose> also idk if that works with global config
2023-01-14 11:53:47 <psykose> and for local there are a bunch of repos
2023-01-14 11:55:16 <ikke> Yeah, and I think it chokes on bb cron
2023-01-14 11:55:19 <ikke> I sent a patch once
2023-01-14 11:55:32 <ikke> bb does something non-standard
2023-01-14 11:56:00 <ikke> git config should not matter if it's global or lcoal
2023-01-14 11:56:03 <ikke> local
2023-01-14 11:56:05 <ikke> it's all merged down
2023-01-14 11:58:08 <psykose> yeah, but i mean for cron, the thing it puts is...
2023-01-14 11:58:09 <psykose> weird
2023-01-14 11:58:16 <psykose> also mine is like completely broken
2023-01-14 11:58:23 <psykose> https://img.ayaya.dev/tZfRCf9wgobu
2023-01-14 11:58:25 <psykose> what the hell is this
2023-01-14 11:59:32 <ikke> psykose: what'
2023-01-14 11:59:37 <ikke> what's wrong with that?
2023-01-14 11:59:47 <ikke> defining a cronjob that is not sensitive to environment
2023-01-14 12:00:14 <psykose> i don't think aports/src/git is an executable
2023-01-14 12:00:27 <psykose> how strange
2023-01-14 12:00:47 <ikke> https://tpaste.us/DBKO that's what it does for me
2023-01-14 12:00:53 <ikke> but it has been a while ago since I set it up
2023-01-14 12:00:57 <psykose> yeah
2023-01-14 12:00:57 <ikke> (the logging I added manually)
2023-01-14 12:01:02 <psykose> curious why it broke for me
2023-01-14 12:01:16 <ikke> Did you by chance install a locally built git or something like that?
2023-01-14 12:01:21 <ikke> what does git --exec-path return?
2023-01-14 12:01:46 <psykose> same as you
2023-01-14 12:01:55 <psykose> (that has been there for a while, idk when it added it)
2023-01-14 12:02:07 <psykose> also doing register now doesn't edit the crontab at all
2023-01-14 12:03:17 <psykose> ah, only on maint start
2023-01-14 12:03:21 <psykose> alright, works i suppose
2023-01-14 12:03:44 <psykose> not sure how to set up cron in a container though except for more containers and mounting
2023-01-14 12:06:26 <ikke> yeah, would require a dedicated container
2023-01-14 12:06:52 <ikke> at least, without involving some kind of service manager
2023-01-14 12:07:11 <ikke> So depends on whether it actually adds something
2023-01-14 12:13:15 <psykose> meh
2023-01-14 12:13:21 <psykose> it's just fetching from gitlab isn't it
2023-01-14 12:13:39 <psykose> hm, though i guess it's distinct anyway
2023-01-14 20:51:35 <psykose> ikke: you still have to kick armv7 :p
2023-01-14 20:52:07 <ikke> Do I have to?
2023-01-14 20:53:54 <ikke> alright alright
2023-01-14 20:55:40 <psykose> it's been 2 days!
2023-01-14 20:55:42 <psykose> :D
2023-01-17 17:31:25 <psykose> ikke: the x86 ci seems to always fail
2023-01-17 17:31:31 <psykose> e.g. https://gitlab.alpinelinux.org/J0WI/aports/-/jobs/951752
2023-01-17 17:31:53 <ikke> hmm, not a lot of information
2023-01-17 17:32:07 <ikke> ofcourse
2023-01-17 17:32:09 <ikke> full disk
2023-01-17 17:42:17 <ikke> lesigh
2023-01-17 17:42:42 <ikke> some user thinking CI is for hosting software in the cloud
2023-01-17 17:59:11 <psykose> haha link
2023-01-17 17:59:13 <psykose> where was it
2023-01-18 05:36:32 <ikke> gitlab security release due to git
2023-01-18 05:48:48 <ikke> fun, it does not build due to missing tag
2023-01-18 10:26:17 <psykose> :)
2023-01-18 10:26:20 <psykose> every time
2023-01-18 11:24:20 <ikke> Ok seems like they have pushed the tags now
2023-01-18 11:50:18 <ikke> yes, build succeeded now
2023-01-18 12:08:23 <psykose> that's good
2023-01-19 12:05:30 <psykose> Free space: 12.9G on the x86_64 ci
2023-01-19 12:39:03 <ikke> added chron to clean up docker
2023-01-19 12:53:43 <ikke> I suppose I'll upgrade gitlab in a bit
2023-01-20 20:45:20 <mps> ikke: clandmeter: idk what you do with gitlab batteries but you made very good 'machine', and I'm proud that it was my proposal and insisting to use gitlab instead of git hub
2023-01-20 20:45:30 <mps> very good job
2023-01-20 20:46:12 <ikke> thank you
2023-01-20 20:46:51 <mps> you deserve big thanks
2023-01-20 20:48:41 <mps> if I forgot to mention psykose then sorry, don't know who works on all these things
2023-01-20 20:49:51 <ikke> clandmeter did most of the work setting it up :)
2023-01-20 20:51:05 <mps> yes, I know this
2023-01-20 20:51:52 <mps> he deserves few bottles of good red wine, maybe full box ;)
2023-01-20 21:24:02 <mps> hm, now when ncopa make linux-lts from the latest stable maybe we don't need linux-edge
2023-01-20 21:57:53 <pj[m]> it's lts
2023-01-20 21:58:49 <mps> pj[m]: no, it is not long term stable
2023-01-20 22:03:13 <psykose> it was upgraded in anticipation of 6.1 being the next lts, early for convenience and all the 6.1 niceties
2023-01-20 22:03:27 <psykose> if it becomes lts then it not currently being lts is just splitting hairs and being pedantic
2023-01-20 22:03:36 <psykose> if it does not then it's just an honest mistake and 6.2 will be the lts
2023-01-20 22:04:29 <psykose> it won't follow latest-stable like edge does
2023-01-20 22:04:44 <mps> wouldn't be easy to to keep it 2 years if it wouldn't be declared lts
2023-01-20 22:05:12 <psykose> obviously, but it's in edge a long time from release and regardless of which gets picked it's in time for next release
2023-01-20 22:06:35 <mps> well, I also expected it will be next lts
2023-01-22 14:03:38 <clandmeter> new gitlab release
2023-01-22 14:04:11 <ikke> 15.8?
2023-01-22 14:07:07 <ikke> yup, 15.8, just received the e-maisl
2023-01-22 14:07:09 <ikke> email
2023-01-22 14:07:58 <ikke> I'm a bit behind atm
2023-01-22 14:08:09 <ikke> I suppose we could try jumping from 15.5 to 15.7
2023-01-22 14:25:26 <ikke> Removing last ruby code from gitaly: https://gitlab.com/gitlab-org/gitaly/-/issues/4742
2023-01-22 14:27:45 <ikke> It still amazes me that the ruby compnents take the longest time to build
2023-01-22 14:56:33 <ikke> upgraded gitlab-test to 15.6
2023-01-22 15:55:34 <ikke> upgraded it now to 15.7
2023-01-22 20:50:32 <psykose> poggers
2023-01-22 20:51:41 <ikke> tried to build 15.8 with alpine 3.17 / ruby 3.1, but 1) it fails to build some gems
2023-01-22 20:53:02 <psykose> what errors
2023-01-22 20:53:57 <ikke> This is one: https://tpaste.us/JRWl 
2023-01-22 20:55:03 <psykose> do you have pkgconf by chance
2023-01-22 20:55:05 <ikke> This is another
2023-01-22 20:55:07 <ikke> https://tpaste.us/8Mok 
2023-01-22 20:55:08 <psykose> also there is more log in mkmf.log
2023-01-22 20:55:10 <ikke> pkgconf is installed
2023-01-22 20:56:05 <psykose> also those "provided configuration options" look funny
2023-01-22 20:56:12 <psykose> it's just --with --without of everything
2023-01-22 20:56:16 <ikke> yeah, noticed
2023-01-22 20:56:34 <ikke> Not something we procide
2023-01-22 20:56:36 <ikke> provide*
2023-01-22 20:56:55 <ikke> Takes a bit more effort to get those log files
2023-01-22 21:00:44 <ikke> dl-master will but unavailable for ~1h tuesday night
2023-01-22 21:01:16 <psykose> sounds ok
2023-01-22 21:01:42 <ikke> host will be upgraded
2023-01-22 21:01:57 <ikke> (kernel)
2023-01-23 11:49:10 <mps> https://sourceware.org/bugzilla/show_bug.cgi?id=30033 bloat of arm32 with binutils bug
2023-01-23 11:50:52 <mps> do we need to revert this for alpine
2023-01-23 11:54:11 <ikke> mps: maybe create an issue for it on gitlab?
2023-01-23 11:54:41 <mps> maybe
2023-01-23 11:54:47 <ikke> why not?
2023-01-23 11:55:40 <mps> I see that qemu-system-arm stopped to work for me on aarch64 but have to investigate reason more deeply
2023-01-23 11:57:40 <mps> ok, qemu-user works
2023-01-23 12:37:23 <psykose> probably fine to wait for a fix
2023-01-23 12:40:15 <psykose> it only affects 32-bit userspace on 64-bit which isn't exactly common (i know you do it) as well as requiring >4k pages (so very few hpc machines, and the m1 ones?)
2023-01-23 12:40:28 <psykose> i.e. the armv7 dev container is 32-bit userspace on 64 and isn't affected on anything i just tested
2023-01-23 12:41:07 <psykose> i know the page size is compile time set for the kernel but
2023-01-23 12:44:48 <mps> m1 uses 16K pages
2023-01-23 12:59:34 <psykose> i know
2023-01-24 17:17:26 <ncopa> m1 does not support 32bit userspace?
2023-01-24 17:17:37 <ikke> no, afaik it does not
2023-01-24 17:17:41 <ikke> just 64-bits
2023-01-24 17:18:10 <psykose> yeah, doesn't
2023-01-24 17:18:19 <psykose> 64-only cpu, like the new iphones afaik
2023-01-24 17:18:38 <ikke> Apple can do that, just discard an entire category of userspace
2023-01-24 17:21:44 <psykose> the tired developer in me wishes more people did the same
2023-01-24 17:22:05 <psykose> people take backwards compat to very absurd lengths sometimes
2023-01-24 17:23:59 <ikke> Not everyone can afford the latest and greatest every other year
2023-01-24 18:11:13 <mps> this way apple packed more 'transistors' on chip
2023-01-24 18:12:06 <mps> but iritates me because I can't run amrv7 lxc on m1. and apple cares for me so much :D
2023-01-24 18:12:56 <mps> ofc, I can but is slow and I will prepare old chromebook to compile armv7
2023-01-24 20:17:28 <mps> psykose: https://tpaste.us/9MD4
2023-01-24 20:17:44 <mps> backtrace from iwd
2023-01-24 20:17:55 <psykose> ah
2023-01-24 20:18:05 <psykose> still looks like missing symbols
2023-01-24 20:18:26 <psykose> i.e. no musl-dbg
2023-01-24 20:18:34 <psykose> but given the error it looks like a stack overrun or whatnot
2023-01-24 20:20:06 <mps> https://tpaste.us/nk1v with musl-dbg
2023-01-24 20:21:05 <psykose> ah
2023-01-24 20:21:07 <psykose> that's the better one
2023-01-24 20:22:01 <psykose> hmm
2023-01-24 20:22:29 <mps> yes, but I don't undestand why it works on kernel 6.1.7 and faults on 6.1.8
2023-01-24 20:23:00 <psykose> the backtrace also indicates it's basically malloc(16)
2023-01-24 20:23:04 <psykose> nothing funny there
2023-01-24 20:24:37 <mps> and my gdb skill is faded away
2023-01-24 20:27:18 <mps> hope upstream dev have deep understandig of this
2023-01-24 21:04:30 <clandmeter> ikke: i think we might need to update the httpd on www
2023-01-24 21:04:40 <ikke> clandmeter: reason?
2023-01-24 21:04:45 <clandmeter> i need to set some headers, but i dont think its possible with darkhttpd
2023-01-24 21:04:49 <ikke> ah ok
2023-01-24 21:05:02 <clandmeter> or i need to do it upstream
2023-01-24 21:05:27 <clandmeter> i also needed to "hack" json support 
2023-01-24 21:05:37 <ikke> Do we want to switch to nginx?
2023-01-24 21:06:02 <clandmeter> i guess that would be the easiest solution
2023-01-24 21:06:15 <clandmeter> currently i get CORS errors
2023-01-24 21:06:32 <clandmeter> and i had to set the default mimetype to json...
2023-01-24 21:06:51 <ikke> I have no issue with switching to nginx
2023-01-24 21:06:51 <clandmeter> err content type
2023-01-24 21:08:23 <mps> clandmeter: when you are here, I reenabled cma for linux-edge on aarch64. it can be disabled with kernel cmdline parameter 'cma=0'
2023-01-24 21:08:54 <clandmeter> ok
2023-01-24 21:09:12 <clandmeter> ill keep it in the back of my mind when we setup the new arm boxes
2023-01-24 21:09:16 <ikke> mps: did something change upstream?
2023-01-24 21:09:28 <clandmeter> i guess somebody had a need for it
2023-01-24 21:09:36 <psykose> darkhttpd has --header doesn't it
2023-01-24 21:09:44 <mps> ikke: no, our users needs cma for some SoCs
2023-01-24 21:09:49 <ikke> mps: ah ok
2023-01-24 21:10:32 <clandmeter> psykose: yes, but you cant set it per path/item
2023-01-24 21:10:39 <psykose> ah
2023-01-24 21:10:39 <clandmeter> its a global thingy
2023-01-24 21:10:40 <psykose> yeah
2023-01-24 21:10:49 <psykose> yeah, nginx is a better choice anyway at this point
2023-01-24 21:10:51 <clandmeter> just like json
2023-01-24 21:11:02 <clandmeter> it can set it based on extension
2023-01-24 21:11:10 <clandmeter> which does not work on things without extension :)
2023-01-24 21:30:19 <clandmeter> I modified the default.conf, should be ok i guess.
2023-01-24 21:30:39 <clandmeter> cors errors gone :)
2023-01-24 21:31:03 <psykose> :D
2023-01-24 21:32:34 <clandmeter> i dont see any performance inpact ;-)
2023-01-24 21:32:39 <clandmeter> impact even
2023-01-24 21:32:59 <psykose> time to add ktls
2023-01-24 21:34:05 <ikke> heh
2023-01-26 08:29:54 <ncopa> good morning! I'd like another wireguard key for a second laptop. how do I add that?
2023-01-26 08:38:06 <mps> ncopa: good morning! same as you made previous one
2023-01-26 08:43:56 <ncopa> i suppose I have to add the pub key to wg.a.o?
2023-01-26 08:44:47 <ncopa> i don't seem to have ssh access to wg.alpinelinux.org
2023-01-26 08:50:00 <mps> ncopa: I see you have it
2023-01-26 08:50:56 <mps> ncopa@ncopa-desktop key
2023-01-26 08:51:29 <ncopa> ok
2023-01-26 08:51:47 <mps> deu7-dev1.alpinelinux.org host
2023-01-26 08:55:50 <ncopa> nice motd message :)
2023-01-26 08:57:44 <mps> oh, that must be ikke ;)
2023-01-26 11:21:08 <ikke> :)
2023-01-26 12:26:50 <PureTryOut> So uh my physical TOTP device is broken and no I can't login to Gitlab anymore. Can I somehow reset the account to get access again?
2023-01-26 12:27:58 <ikke> PureTryOut: let me check
2023-01-26 13:33:30 <clandmeter> PureTryOut: i guess you use it for more than one service. Sounds like fun..
2023-01-26 13:33:46 <PureTryOut> Yup... Great fun indeed
2023-01-26 13:34:11 <PureTryOut> GitHub was quite easy to reset, but I need to switch it on multiple Gitlab instances
2023-01-26 19:45:45 <ikke> psykose: ftr, the host keys are shown here: https://gitlab.alpinelinux.org/help/instance_configuration#ssh-host-keys-fingerprints
2023-01-26 19:46:04 <ikke> or the fingerprints
2023-01-26 19:47:46 <psykose> ay fixed
2023-01-26 22:01:12 <psykose> ikke: i broke rust by accident so i'll need your help to fix it :p
2023-01-26 22:04:41 <psykose> the install got generated with 0600 perms so the package is broken on install
2023-01-26 22:04:51 <psykose> https://git.alpinelinux.org/aports/commit/?id=5c958a29ca22 fixes it, but the existing one can't be used to build it again
2023-01-26 22:05:02 <psykose> so you have to rebuild all of them by hand after running that on the installed rust
2023-01-26 22:05:17 <psykose> unless you have a better idea
2023-01-27 05:13:24 <ikke> Install rust, fix permissions, build rust? 
2023-01-27 05:48:24 <psykose> yeah
2023-01-27 05:48:52 <psykose> abuild deps; run that find -exec in the commit on the same /usr path; abuild -r; repeat for all 8 arches
2023-01-27 05:48:55 <psykose> :/
2023-01-27 05:49:37 <psykose> annoying busywork for you, so apologies
2023-01-27 05:49:45 <psykose> i should've double checked the artifacts
2023-01-27 05:50:04 <psykose> not the first time i saw this issue by chance, i tried debugging it before but couldn't find anything other than "caused by sccache"
2023-01-27 05:50:08 <psykose> and now it happens regardless
2023-01-27 05:50:38 <psykose> trying to debug build steps in rustbuild isn't very easy
2023-01-27 05:51:28 <ikke> It's already building
2023-01-27 05:51:52 <psykose> :)
2023-01-27 05:52:36 <ikke> I just manually installed rust, so need to make sure we remove it afterwards
2023-01-27 05:52:42 <psykose> ah, ok
2023-01-27 05:52:42 <psykose> yeah
2023-01-27 05:52:49 <psykose> if you did just abuild deps it would remove it for you after
2023-01-27 05:53:07 <ikke> yes, but then I need to switch between buildozer and root :)
2023-01-27 05:53:24 <psykose> :p
2023-01-27 05:53:27 <psykose> don't forget
2023-01-27 05:53:38 <psykose> riscv will also take quite a while compared to the others
2023-01-27 05:53:44 <ikke> yup
2023-01-27 06:36:15 <ikke> wow, armhf finished first
2023-01-27 06:37:08 <psykose> :)
2023-01-27 06:37:21 <ikke> fyi, I removed rust from world everywhere
2023-01-27 06:38:49 <psykose> okie
2023-01-27 06:38:49 <ikke> maybe others finished earlier, but they are building other packages
2023-01-27 06:38:53 <psykose> thanks boss
2023-01-27 18:54:18 <ikke> thinking about upgrading gitlab tonight
2023-01-27 19:20:34 <psykose> good idea
2023-01-27 19:20:54 <ikke> from 15.5 to 15.7, tested that on gitlab-test
2023-01-27 19:38:18 <ikke> psykose: let me know if you notice issues. One thing I changed is to enable gitally to used the bundled git binaries (recommended)
2023-01-27 19:51:33 <psykose> will do
2023-01-28 06:42:17 <psykose> ikke: gitlab merges don't trigger builders
2023-01-28 07:18:24 <ikke> psykose: hmm, need to check  the webhook then. 
2023-01-28 07:36:35 <ikke> psykose: see a 500 response in the webhooks (and gitlab started to temporarily suspoend webhooks when that happens)
2023-01-28 07:36:43 <psykose> unfortunate
2023-01-28 07:51:08 <ikke> psykose: /usr/bin/lua5.3: /etc/webhooks/scripts/alpine-disk-image:13: attempt to index a nil value (field 'project')
2023-01-28 07:51:34 <psykose> guess they changed some apis
2023-01-28 07:52:03 <ikke> /srv/compose/webhook-new on gbr2-dev1 in case you have some time (I have time later)
2023-01-28 07:52:59 <ikke> ah, project changed to project_name
2023-01-28 07:54:19 <ikke> seems like it has flattened
2023-01-28 07:56:52 <ikke> ok, get 200 back now
2023-01-28 08:00:12 <ikke> psykose: it was changing a . to a _ :)
2023-01-28 08:01:02 <psykose> :)
2023-01-31 18:26:52 <clandmeter> ikke: im getting a "in your face" critical security release update request on gitlab.
2023-01-31 18:27:05 <clandmeter> i guess you probably seen it already
2023-01-31 18:27:16 <ikke> no, not yet
2023-01-31 18:27:22 <ikke> I guess you are ahead in the queue :)
2023-01-31 18:27:37 <clandmeter> i have never seen this overlay before
2023-01-31 18:27:50 <ikke> Oh, me neither
2023-01-31 18:27:52 <ikke> I see it now
2023-01-31 18:28:05 <clandmeter> makes sense to do it like this
2023-01-31 18:28:27 <ikke> meh
2023-01-31 18:28:39 <ikke> only a couple of medium severity issues
2023-01-31 18:28:54 <ikke> This is a 'stop everything you are doing and upgrade immediately' popup
2023-01-31 18:29:05 <clandmeter> :)
2023-01-31 18:29:12 <clandmeter> yes, i didnt look at the details yet
2023-01-31 18:29:22 <clandmeter> i guess its just a new feature 
2023-01-31 18:29:37 <ikke> "Critical security release"
2023-01-31 18:29:47 <clandmeter> will popup every patch release :p
2023-01-31 18:31:00 <clandmeter> oh it also has a nice header now
2023-01-31 18:31:04 <ikke> yes
2023-01-31 18:31:17 <clandmeter> i pressed Upgrade now, but nothing happend...
2023-01-31 18:31:25 <psykose> >Sidekiq background job DoS
2023-01-31 18:31:36 <psykose> i thought that was just running sidekiq at all? nothing new here
2023-01-31 18:31:56 <ikke> "CSRF via file upload allows an attacker to take over a repository." seems to be the most serious
2023-01-31 18:32:14 <psykose> next to a caveat of "if an Owner or Maintainer uploads a file"
2023-01-31 18:33:33 <psykose> >make a sidekiq job allocate a lot of memory
2023-01-31 18:33:40 <psykose> just so much humor in these notes
2023-01-31 18:33:57 <clandmeter> psykose: i created this issue, not sure you already seen it?  https://gitlab.alpinelinux.org/alpine/tsc/-/issues/65
2023-01-31 18:34:05 <psykose> i did, how come
2023-01-31 18:34:55 <clandmeter> why i created it?
2023-01-31 18:35:03 <psykose> no, like
2023-01-31 18:35:04 <psykose> what about it
2023-01-31 18:36:00 <clandmeter> you are also part of infra, so i thought i would let you know it exists. 
2023-01-31 18:36:29 <clandmeter> although you probably get notified as you are a tsc member now
2023-01-31 18:36:51 <psykose> i get notified on almost everything :)
2023-01-31 18:37:45 <psykose> i don't personally use matrix or care about the proposal
2023-01-31 18:38:07 <clandmeter> nod
2023-01-31 18:38:24 <clandmeter> its not a proposal to push ppl to use matrix
2023-01-31 18:39:22 <ikke> https://gitlab.com/gitlab-org/gitlab/-/issues/387719
2023-01-31 18:43:40 <psykose> it would probably make the most sense to bridge only #alpine-linux and no other channels, given that bridged matrix channels fall into one of two categories
2023-01-31 18:43:54 <psykose> - people know how irc works and the channel on the matrix side is just an irc channel in essence
2023-01-31 18:44:02 <psykose> - people don't, and every channel looks like
2023-01-31 18:44:12 <psykose> * - people dont, and every channel looks like
2023-01-31 18:44:17 <psykose> * * - people dont, and every channel looks lik
2023-01-31 18:44:22 <psykose> etc
2023-01-31 18:45:49 <psykose> the matrix-only channels can do whatever (while being official, etc etc), i'm not going to be there so it's no difference for me
2023-01-31 18:46:03 <psykose> the bridged ones are guaranteed to be a disaster so at least keep it to one :)
2023-01-31 18:46:59 <clandmeter> i was expecting controversy, that's why I would like to discuss it in the TSC.
2023-01-31 18:48:20 <psykose> i don't think it's very controversial, more just the benefit is in having an official space now including another platform
2023-01-31 18:48:38 <psykose> so if it's non-bridged rooms it's just more spaces for whoever wants them (assuming someone can moderate them)
2023-01-31 18:48:53 <psykose> and #alpine-linux could be bridged just for fun as it's a free-for-all anyway
2023-01-31 18:49:07 <psykose> but other channels wouldn't gain anything from it, there's no benefit to e.g. #alpine-devel
2023-01-31 18:49:43 <psykose> none of us use matrix and anyone that does can already join the channel via the automatic compatibility anyway, making it an extra room on the other side doesn't improve what it's for
2023-01-31 18:50:59 <psykose> so in effect the proposal is "making another social media space", which is fine if someone wants to maintain all that, no real issues
2023-01-31 18:55:50 <clandmeter> not sure what the difference is with the current status. alpine-devel is already bridged. what other disaster could happen?
2023-01-31 18:56:03 <psykose> bridged to what
2023-01-31 18:56:14 <clandmeter> bridged to irc
2023-01-31 18:56:29 <psykose> we are on irc, it's bridged to itself?
2023-01-31 18:56:51 <clandmeter> matrix is bridged to alpine-devel irc
2023-01-31 18:56:59 <pj[m]> it's bridged to matrix
2023-01-31 18:57:01 <pj[m]> the other way around, cland
2023-01-31 18:57:19 <clandmeter> a bridge doesnt work two ways?
2023-01-31 18:57:29 <pj[m]> it does, but the irc is origin
2023-01-31 18:57:31 <psykose> i don't mean the automatic "matrix users can join any irc room"
2023-01-31 18:58:02 <psykose> do you have some actual bridge to a room set up or
2023-01-31 18:59:23 <pj[m]> although I agree that having dedicated matrix server will help with moderation
2023-01-31 19:01:53 <psykose> i.e. by default there is a matrix<->oftc bridge, and it permits #_oftc_#channame:matrix.org
2023-01-31 19:02:00 <psykose> and matrix users can join any irc room through that
2023-01-31 19:02:16 <psykose> this is effectively meaningless though, it's just individual users
2023-01-31 19:02:41 <clandmeter> yes you specifically need to join that room, and are aware its a bridged irc room
2023-01-31 19:02:48 <psykose> an actual bridged room would be #alpine-official-whatever:matrix.org, and that room is bridged into some equivalent irc room
2023-01-31 19:02:59 <psykose> and it would be a room advertised as a matrix room, etc
2023-01-31 19:03:06 <pj[m]> You can make a matrix space if you need to have them organised
2023-01-31 19:03:34 <pj[m]> but other than that and moderation I don't see much point
2023-01-31 19:03:57 <clandmeter> if you link the irc channel to the matrix room, users that would join the room would automatically also join irc.
2023-01-31 19:04:00 <clandmeter> iiuc
2023-01-31 19:04:06 <pj[m]> this already happens
2023-01-31 19:04:12 <pj[m]> right now, for me
2023-01-31 19:04:25 <clandmeter> which room did you join?
2023-01-31 19:04:41 <pj[m]> I'm on #oftc_alpine-infra or something like that
2023-01-31 19:04:45 <psykose> the distinction is that #_oftc_#whatever is clear that it's bridged and only people that use a matrix client to access irc rooms use it
2023-01-31 19:04:57 <clandmeter> yes thats what i said
2023-01-31 19:05:15 <psykose> if you have an official matrix room and it's bridged, people join it just to use the matrix room, they don't necessarily care it's two places at once
2023-01-31 19:05:19 <clandmeter> if you link the room, then you join @devel:alpinelinux.org and automatically also join irc.
2023-01-31 19:05:44 <pj[m]> I don't want to sound rude, but "ok but why?"
2023-01-31 19:06:05 <pj[m]> as psykose said, everything is already linekd
2023-01-31 19:06:07 <psykose> and then, being unaware that it's actually bridged, they use it like a matrix room
2023-01-31 19:06:11 <pj[m]> linked, bleh
2023-01-31 19:06:19 <psykose> 5 minutes later the entire chat is urls and edited messages 
2023-01-31 19:06:33 <psykose> so.. it's just a matrix room and unreadable garbage on the irc side
2023-01-31 19:06:59 <pj[m]> being mostly an IRC user here and elsewhere, I've felt pain of people who edit same message >5 times
2023-01-31 19:07:14 <pj[m]> and it's extremely annoying when someone writes essays
2023-01-31 19:07:57 <clandmeter> there is no difference now, all those matrix users are already here. 
2023-01-31 19:08:06 <psykose> yes, and they are aware it's bridged
2023-01-31 19:08:10 <psykose> i feel like you didn't read anything i said
2023-01-31 19:08:26 <clandmeter> same goes for me
2023-01-31 19:08:33 <psykose> ok
2023-01-31 19:10:40 <pj[m]> > bridge irc<>matrix | already done
2023-01-31 19:10:42 <pj[m]> > joins irc when joined matrix | already done
2023-01-31 19:11:25 <pj[m]> > know that it's bridge | already done (can be seen by channel name)
2023-01-31 19:12:39 <pj[m]> the only thing that can't be done is vanity matrix home server (here is :matrix.org everywhere vs :alpinelinux.org) and I believe moderation also slightly sucks
2023-01-31 19:13:13 <pj[m]> (because you don't manage :matrix.org so don't have as high perms and network bans are not possible)
2023-01-31 19:15:06 <pj[m]> although moderation can be improved by setting the channel to registered only so people have to link their IRC with matrix
2023-01-31 19:17:08 <pj[m]> but also mind that moderation tools on matrix side also suck, because there is matrix API and ??? eh, mjolnir? which you also have to host yourself and is massive PITA
2023-01-31 19:19:34 <psykose> clandmeter: but yeah, i don't mind (as i don't use matrix, so it's not up to me of course). you can implement it as you wish, i don't mind :) no worries
2023-01-31 19:20:12 <pj[m]> one other positive effect is that if you will host bridge, no more oftc irc<>matrix bridge outage?
2023-01-31 19:20:42 <pj[m]> and randomly kicking out everyone and bridge being dead for days
2023-01-31 19:20:53 <clandmeter> psykose: its not my wish, my wish is to discuss and see the cons/benefits.
2023-01-31 19:21:27 <clandmeter> and if there is too much cons, or there is too much controversy, we just dont support it.
2023-01-31 19:21:49 <psykose> sure, guess we'll discuss it in the meeting officially then
2023-01-31 19:22:16 <clandmeter> pj[m]: you are a irc user or a matrix user?
2023-01-31 19:22:19 <ikke> How will this affect the users who are now using the existing bridge?
2023-01-31 19:22:33 <clandmeter> nothing i think
2023-01-31 19:22:49 <psykose> slightly tangential and off topic, but all the "unofficial" alpine linux spaces on various social media i've come across were trainwreck disasters
2023-01-31 19:22:50 <pj[m]> clandmeter: I'm using just my matrix account right now because I'm too lazy to connect it to IRC and I had some issues long time ago
2023-01-31 19:23:03 <psykose> i'd guess having an official one in something more friendly than irc (non-bridged) might be useful in that regard
2023-01-31 19:23:18 <pj[m]> connect as in "link my matrix account via IRC bridge to my IRC account"
2023-01-31 19:23:30 <clandmeter> and no i dont think we will run our own bridge
2023-01-31 19:23:31 <pj[m]> but I talk often from IRC too
2023-01-31 19:26:15 <clandmeter> i think its a valid point that ppl that are unaware it is bridged could create issues for irc users.
2023-01-31 19:27:34 <clandmeter> but i have not noticed it that much that it an issue for me
2023-01-31 19:28:02 <psykose> it's not an issue on any of our channels for sure
2023-01-31 19:28:10 <ikke> It's better now, but we regular had to tell people how the features they are using appear on irc
2023-01-31 19:28:23 <pj> what could help is set topic for matrix channels
2023-01-31 19:29:44 <pj> nevermind, they are set, cinny just doesn't show them next channel name
2023-01-31 19:29:51 <clandmeter> ikke: true, but we have told as many not to paste into irc 
2023-01-31 19:39:10 <clandmeter> i remember when i joined one of the pmos rooms, it asked me to confirm the room rules or whatever to enter.
2023-01-31 19:39:47 <clandmeter> but i guess everybody just pushes agree and carries on :)