<@sgallagh:fedora.im>
17:00:58
!startmeeting FESCO (2024-11-19)
<@meetbot:fedora.im>
17:01:00
Meeting started at 2024-11-19 17:00:58 UTC
<@meetbot:fedora.im>
17:01:01
The Meeting name is 'FESCO (2024-11-19)'
<@sgallagh:fedora.im>
17:01:05
!meetingname fesco
<@meetbot:fedora.im>
17:01:06
The Meeting Name is now fesco
<@sgallagh:fedora.im>
17:01:11
Chairs: @conan_kudo:matrix.org, @ngompa:fedora.im, @nirik:matrix.scrye.com, @humaton:fedora.im, @zbyszek:fedora.im, @sgallagh:fedora.im, @jistone:fedora.im, @dcantrell:fedora.im, @decathorpe:fedora.im, @salimma:fedora.im
<@sgallagh:fedora.im>
17:01:16
!topic Init Process
<@sgallagh:fedora.im>
17:01:17
!hi
<@zodbot:fedora.im>
17:01:19
Stephen Gallagher (sgallagh) - he / him / his
<@dcantrell:fedora.im>
17:01:28
!hi
<@zodbot:fedora.im>
17:01:32
David Cantrell (dcantrell) - he / him / his
<@jistone:fedora.im>
17:01:36
!hi
<@zodbot:fedora.im>
17:01:38
Josh Stone (jistone) - he / him / his
<@salimma:fedora.im>
17:01:40
!hi
<@zodbot:fedora.im>
17:01:41
Michel Lind (salimma) - he / him / his
<@nirik:matrix.scrye.com>
17:02:36
morning
<@zbyszek:fedora.im>
17:02:38
!hi
<@zodbot:fedora.im>
17:02:39
Zbigniew Jędrzejewski-Szmek (zbyszek)
<@sgallagh:fedora.im>
17:03:08
That's quorum, so I guess we'll begin
<@zbyszek:fedora.im>
17:03:30
"fun"
<@conan_kudo:matrix.org>
17:03:51
!hi
<@zodbot:fedora.im>
17:03:54
Neal Gompa (ngompa) - he / him / his
<@salimma:fedora.im>
17:03:55
I thought this meeting is where fun goes to die
<@salimma:fedora.im>
17:04:00
oh so maybe that's appropriate :P
<@sgallagh:fedora.im>
17:04:03
!topic #3268 Nonresponsive maintainer: Fabian Affolter fab
<@sgallagh:fedora.im>
17:04:06
!fesco 3268
<@zodbot:fedora.im>
17:04:06
● **Last Updated:** 12 hours ago
<@zodbot:fedora.im>
17:04:06
● **Assignee:** Not Assigned
<@zodbot:fedora.im>
17:04:06
**fesco #3268** (https://pagure.io/fesco/issue/3268):**Nonresponsive maintainer: Fabian Affolter fab**
<@zodbot:fedora.im>
17:04:06
<@zodbot:fedora.im>
17:04:06
● **Opened:** 2 months ago by fale
<@nirik:matrix.scrye.com>
17:05:07
I've not had a chance to look at the PR yet (pesky sleep)
<@sgallagh:fedora.im>
17:05:10
zbyszek: You added the meeting tag. I assume you want to discuss the new "lite" nonresponsive policy?
<@sgallagh:fedora.im>
17:05:43
and/or Michel Lind 🎩 UTC-6
<@zbyszek:fedora.im>
17:05:43
That was for last week.
<@zbyszek:fedora.im>
17:05:56
I guess I must've forgotten to unset it.
<@decathorpe:fedora.im>
17:06:05
!hi
<@zodbot:fedora.im>
17:06:06
Fabio Valentini (decathorpe) - he / him / his
<@salimma:fedora.im>
17:06:12
I do have a PR I just put up that I want some discussion on
<@salimma:fedora.im>
17:06:46
it can be now or at open floor - I have a feeling we want to make changes to the initial proposal, but I tried to keep it as close to EPEL first and we can discuss what changes are needed
<@humaton:fedora.im>
17:07:21
!hi
<@zbyszek:fedora.im>
17:07:22
There has been a bunch of comments on the pull request. Will you be putting up a new version?
<@zodbot:fedora.im>
17:07:23
Tomáš Hrčka (humaton) - he / him / his
<@nirik:matrix.scrye.com>
17:08:00
also perhaps this should be discussed more widely before implementing anything?
<@salimma:fedora.im>
17:08:01
yeah, I'll revise after this meeting
<@sgallagh:fedora.im>
17:08:14
nirik: I was about to suggest the same thing
<@sgallagh:fedora.im>
17:08:29
Maybe we can use this time to hammer out what we (FESCo) think the policy should be?
<@nirik:matrix.scrye.com>
17:08:37
needinfo is pretty anoying, but we do use it in other places too...
<@salimma:fedora.im>
17:08:49
<@salimma:fedora.im>
17:08:49
one question for everyone though
<@salimma:fedora.im>
17:08:49
should we require fesco review the MR? so instead of asking releng we say "after X weeks ask FESCo to review and grant you comaintainership"
<@zbyszek:fedora.im>
17:09:14
I like this idea.
<@salimma:fedora.im>
17:09:23
agreed that 4 weeks is too long for me too, someone suggested it here last week so I thought I'll write it and see what people think
<@sgallagh:fedora.im>
17:09:39
Yes, but perhaps not "FESCo as a whole" but, "At least one member of FESCo"?
<@salimma:fedora.im>
17:10:08
<@salimma:fedora.im>
17:10:08
so... get fesco to review after 3 weeks, it can be a normal voting process I suppose after that?
<@salimma:fedora.im>
17:10:08
and I'll make it so a PR is required, needinfo is after a week, and document this on top of the page as we want people to use this first
<@sgallagh:fedora.im>
17:10:11
(trusting in that FESCo member to escalate to the full committee if needed)
<@salimma:fedora.im>
17:10:13
and remove any mention of 'trial'
<@nirik:matrix.scrye.com>
17:10:26
I worry we will hit cases where maintainers are super busy/away and someone just get added and messes up something. But of course the other side is that we need to keep things moving.
<@salimma:fedora.im>
17:10:57
oh nice, I like that. so.. file fesco issue, and any member that want to process this can do so, and if someone sees a red flag, vote it down and we have a meeting?
<@nirik:matrix.scrye.com>
17:11:06
and adding that in makes it less 'light weight' 😄
<@sgallagh:fedora.im>
17:11:11
nirik: That can happen with any comaintainer. I'd rather we err on the side of faster progress, personally
<@zbyszek:fedora.im>
17:11:20
I think we should use the same voting rules as for the non-responsive maintainer policy. "If at least one FESCo member votes +1 and no one votes differently, the ticket is approved after three days. If any -1 or 0 votes are made, FESCo will discuss the issue during a meeting."
<@sgallagh:fedora.im>
17:11:26
Michel Lind 🎩 UTC-6: My thoughts exactly
<@nirik:matrix.scrye.com>
17:11:32
note that you can't needinfo all comaintainers, only a specific account
<@salimma:fedora.im>
17:11:38
also... question re: how the old non-responsive maintainer process works. if we agree to rescind packaging rights from someone ... how do we even do it? do we ask releng to? (I personally don't think I have the right to edit ACLs)
<@salimma:fedora.im>
17:11:59
that seems lightweight enough
<@nirik:matrix.scrye.com>
17:12:09
the old process doesn't remove them from packager, just from the packages.
<@salimma:fedora.im>
17:12:15
you can but you need to do it one by one :(
<@salimma:fedora.im>
17:12:36
at least... I think you can. I have seen packages with multiple needinfo, but not more than 2, so idk if there is a limit
<@nirik:matrix.scrye.com>
17:12:39
and you would need to look those up... but ok, sure.
<@nirik:matrix.scrye.com>
17:12:52
I had some packages with double needinfo to the same account.
<@salimma:fedora.im>
17:12:56
ah, sorry, yes. still, do one of us do it normally or do we just ask releng?
<@sgallagh:fedora.im>
17:13:13
Michel Lind 🎩 UTC-6: Usually it's nirik who wears both hats
<@nirik:matrix.scrye.com>
17:13:30
releng... unless you have acls for the package I guess?
<@salimma:fedora.im>
17:14:28
even as an admin you can't kick out the main admin I thought
<@nirik:matrix.scrye.com>
17:14:39
lots of people also filter bugzilla emails... thats why the non responsive process had a devel list post/asking for how to contact them
<@salimma:fedora.im>
17:14:41
but anyway, I guess that's internal book keeping and doesn't need to be in the document
<@nirik:matrix.scrye.com>
17:15:09
there's a 'give package/project' thing... not sure if it's restricted to main admin or any admin
<@humaton:fedora.im>
17:15:21
to main admin and releng
<@sgallagh:fedora.im>
17:15:58
Point of order: let's not dive too deep into the weeds of how Pagure works, since we expect to be migrating off of it once the Forge replacement arrives
<@salimma:fedora.im>
17:16:01
<@salimma:fedora.im>
17:16:01
we don't need to tell people to email devel for the lightweight process right? that makes it.. less lightweight
<@salimma:fedora.im>
17:16:01
needinfo seems to normally be noticed, esp since now bugzilla helpfully flag up how many needinfo you have open
<@salimma:fedora.im>
17:16:08
+1
<@salimma:fedora.im>
17:16:43
ok, I'll revise the PR, I think I have everything I need (unless someone wants to suggest further changes)
<@conan_kudo:matrix.org>
17:18:40
note that there are large classes of maintainers that blackhole BZ emails
<@conan_kudo:matrix.org>
17:19:12
Also, I would prefer this process not require a month and change to execute on
<@conan_kudo:matrix.org>
17:19:21
that makes it longer than even our unresponsive maintainer process
<@zbyszek:fedora.im>
17:19:42
If maintainers blackhole emails, then they're not good maintainers. How are people supposed to interact with them?
<@salimma:fedora.im>
17:20:00
so far it's 1 week + 2 weeks needinfo + 3 days?
<@sgallagh:fedora.im>
17:20:03
Carrier pigeons. Clearly.
<@conan_kudo:matrix.org>
17:20:05
I agree, but that doesn't change that it happens and we don't really have a recourse for when they do
<@conan_kudo:matrix.org>
17:20:23
I'm aware of two very large groups of maintainers that do this on purpose
<@nirik:matrix.scrye.com>
17:20:24
There are many workflows. ;) I know some packages that drop all bugzilla emails and just use queries to get bugs when they triage/work them
<@jistone:fedora.im>
17:20:38
policy shouldn't *cater* to that case though, even if we're aware of it
<@jistone:fedora.im>
17:21:15
(queries are fine, I mean outright ignoring)
<@zbyszek:fedora.im>
17:22:04
<@zbyszek:fedora.im>
17:22:04
I think Neal has a valid point that this proposed process is too lengthy. I think we should make it more like the current nonresp-maintainer process. C.f.: "Week 1. If approved, and the reporter is a current Fedora packager in good standing, interested in comaintaining the package, FESCo will default to adding the reporter as the package admin."
<@zbyszek:fedora.im>
17:22:04
So that means that you can become comaintainer after ~10 days from the start of the process.
<@nirik:matrix.scrye.com>
17:22:44
that does make it easier/faster to add co-maintainers... but of course many people go on vacations longer then 10 days.
<@zbyszek:fedora.im>
17:22:45
So in the new lightweight process, I'd propose the same. In Week 1., after fesco approves, the ticketer becomes a package maintainer.
<@conan_kudo:matrix.org>
17:23:05
yeah, it's not supposed to be the end of the world when there are new maintainers
<@salimma:fedora.im>
17:23:15
nice
<@zbyszek:fedora.im>
17:23:31
That's why the nonresp-maintainer process checks vacation status as the very first step.
<@nirik:matrix.scrye.com>
17:23:46
sure, if folks update that status. ;)
<@salimma:fedora.im>
17:23:54
wait, so escalate to FESCO after only a week?
<@conan_kudo:matrix.org>
17:23:54
we should probably tell people about that system
<@decathorpe:fedora.im>
17:24:04
so ... this new lightweight process is basically a subset of the existing non-responsive maintainer procedure?
<@salimma:fedora.im>
17:24:08
yeah... almost nobody uses the vacation calendar
<@conan_kudo:matrix.org>
17:24:30
to be fair, I didn't even know it existed for quite a long time
<@nirik:matrix.scrye.com>
17:24:42
Communication is hard.
<@salimma:fedora.im>
17:24:43
there was a tool that used to query it but it broke
<@zbyszek:fedora.im>
17:24:46
Dunno, I use the vacation calendar, and whenever I go to fill it in, there's always a bunch of entries.
<@sgallagh:fedora.im>
17:24:56
This isn't revoking their permissions, it's granting it to someone who is active
<@zbyszek:fedora.im>
17:25:01
And I see lots of people with "away" mentions in their matrix handles.
<@zbyszek:fedora.im>
17:25:02
Eetc.
<@sgallagh:fedora.im>
17:25:12
I think that's perfectly reasonable to do on a shorter timeframe, even if the maintainer is away
<@decathorpe:fedora.im>
17:25:22
was this in response to me?
<@decathorpe:fedora.im>
17:25:22
that's why I said *subset* ...
<@sgallagh:fedora.im>
17:25:26
Arguably, that makes it MORE important to add a comaintainer
<@salimma:fedora.im>
17:25:37
file bug (must have a PR), after a week, NEEDINFO but immediately file with FESCo so the maintainer has 3 days basically to respond or one of us get releng to grant the requester admin rights?
<@conan_kudo:matrix.org>
17:25:39
right
<@zbyszek:fedora.im>
17:25:53
If we accept removal of all ownership in a given timeframe, I think we should also accept adding comaintainers.
<@sgallagh:fedora.im>
17:26:01
Fabio Valentini 🤒: No, not in response to you. In response to "people go on vacations"
<@salimma:fedora.im>
17:26:05
yeah, that would speed up a lot of things
<@salimma:fedora.im>
17:26:05
it's also much faster than the EPEL process, so I wonder if that should be merged into this somehow at some point
<@salimma:fedora.im>
17:26:05
<@sgallagh:fedora.im>
17:26:08
I think this *helps* that situation, honestly
<@conan_kudo:matrix.org>
17:26:54
we need to make more headway in the concept that people don't own packages
<@salimma:fedora.im>
17:26:57
fair. though I thought the non-responsive one starts *after* the requester already waited a while on a bug. so the 2 weeks or so is not from the first time they wanted a package changed/fixed
<@salimma:fedora.im>
17:27:11
but I'm personally happy to see this work in 10 days total
<@conan_kudo:matrix.org>
17:27:13
one of the reasons we changed the words in pagure and in the email aliases was to help that along
<@nirik:matrix.scrye.com>
17:27:15
The case I worry about: someone is working painfully on some upgrade, they almost have things lined up. They go to take a break over a weekend and when they come back, someone has been added as comaintainer and pushed the update and messed up everything. ;)
<@zbyszek:fedora.im>
17:27:49
We have had the general idea that packages should have multiple maintainers as a goal since forever. If we make adding comaintainers easier, that'll help.
<@salimma:fedora.im>
17:27:49
oh yeah, this will be faster than the EPEL process, has the full blessing of FESCo, so ... if it works out we can even just drop the EPEL part or align it with this
<@sgallagh:fedora.im>
17:27:51
I think I've proposed in the past that we remove packagers from primary maintainer whenever they complain that they "own" the package and someone else changed it.
<@nirik:matrix.scrye.com>
17:27:53
I guess the best way to avoid that kind of thing is to have a bug and keep status and hope any drive by people will see it?
<@decathorpe:fedora.im>
17:28:02
I think what I'm trying to say is - does this need to be a separate process?
<@decathorpe:fedora.im>
17:28:02
or could we amend the non-responsive process to have two steps, with the second one (getting ACLs revoked on all packages) being optional?
<@salimma:fedora.im>
17:28:21
I think that's the end goal
<@sgallagh:fedora.im>
17:28:27
nirik: Honestly, that's what a draft MR is for
<@sgallagh:fedora.im>
17:28:45
You indicate that there's work in progress, but it's not done and someone else could possibly finish it
<@salimma:fedora.im>
17:28:47
but do we want to try the new process first, and then do a second pass to clean up the old process, or do it all at once?
<@salimma:fedora.im>
17:28:56
(happy to do whichever people prefer)
<@nirik:matrix.scrye.com>
17:29:02
does the process have you check other open bugs and prs?
<@decathorpe:fedora.im>
17:29:06
I think both is fine, but I was just confused :)
<@sgallagh:fedora.im>
17:29:14
782357
<@sgallagh:fedora.im>
17:29:26
Ignore that. Bumped my yubikey
<@salimma:fedora.im>
17:29:45
your yubikey produce 6-digit TOTP codes?
<@sgallagh:fedora.im>
17:30:10
Off-topic
<@salimma:fedora.im>
17:30:39
for the lightweight process I think we should not
<@salimma:fedora.im>
17:31:02
for the full revoking from all packages, the submitter is supposed to document that, right? I think we should keep that part
<@nirik:matrix.scrye.com>
17:31:05
but if you are asking to be a co-maintainer, shouldn't you know what work is in progress?
<@decathorpe:fedora.im>
17:31:19
I think that requirement should remain
<@sgallagh:fedora.im>
17:31:38
We are now at 30 minutes on this topic, do we want to continue discussing here or take it back to the ticket to formalize a proposal?
<@salimma:fedora.im>
17:31:40
oh fair
<@salimma:fedora.im>
17:31:45
but limited to the package in question
<@zbyszek:fedora.im>
17:31:50
I think the requirement should remain, but just scoped to the package.
<@zbyszek:fedora.im>
17:31:59
I.e. check activity on _that package_.
<@salimma:fedora.im>
17:32:02
whereas for general non-responsive you can cite examples from multiple packages and that strengthens the case
<@nirik:matrix.scrye.com>
17:32:05
yeah, I wasn't suggesting it should be all packages
<@salimma:fedora.im>
17:32:14
yup
<@salimma:fedora.im>
17:32:22
ok, I'll make sure the revised PR makes that clear
<@nirik:matrix.scrye.com>
17:32:24
anyhow, I think this needs a bunch more discussion (hopefully on devel/discussion?)
<@salimma:fedora.im>
17:32:52
yeah. I'll revise the PR, give people time to look, and then send it to devel/discussion?
<@salimma:fedora.im>
17:33:04
and do a separate PR to fix the antora build that we should land faster :)
<@zbyszek:fedora.im>
17:33:15
Sounds good.
<@nirik:matrix.scrye.com>
17:33:37
excellent!
<@sgallagh:fedora.im>
17:33:48
Thanks, Michel Lind 🎩 UTC-6
<@zodbot:fedora.im>
17:33:55
sgallagh gave a cookie to salimma. They now have 51 cookies, 5 of which were obtained in the Fedora 41 release cycle
<@zodbot:fedora.im>
17:34:03
decathorpe gave a cookie to salimma. They now have 52 cookies, 6 of which were obtained in the Fedora 41 release cycle
<@zodbot:fedora.im>
17:34:10
humaton has already given cookies to salimma during the F41 timeframe
<@sgallagh:fedora.im>
17:34:21
!action Michel Lind 🎩 UTC-6 to revise the proposal and send it to a wider audience for further discussion
<@sgallagh:fedora.im>
17:34:53
That was the only pre-announced topic, so we'll select the next Chair and then move to nirik 's Open Floor topics.
<@sgallagh:fedora.im>
17:34:58
!topic Next Week's Chair
<@zodbot:fedora.im>
17:34:59
kevin has already given cookies to salimma during the F41 timeframe
<@sgallagh:fedora.im>
17:35:14
I'll be away next week for the US Thanksgiving holiday.
<@nirik:matrix.scrye.com>
17:35:21
I will not be around next week. :) (us thanksgiving)
<@dcantrell:fedora.im>
17:35:33
same, I will not be around for thanksgiving
<@jistone:fedora.im>
17:36:18
I can do it, if we manage quorum
<@zbyszek:fedora.im>
17:36:26
Conan Kudo what about you?
<@humaton:fedora.im>
17:36:33
I will be around
<@salimma:fedora.im>
17:36:34
I'll be around, though might be distracted as the kid's daycare is closed all week
<@conan_kudo:matrix.org>
17:36:42
I will be somewhat around
<@conan_kudo:matrix.org>
17:36:47
staycationing but around
<@nirik:matrix.scrye.com>
17:36:55
if we don't have anything urgent we could just cancel?
<@conan_kudo:matrix.org>
17:37:02
I'm fine with that
<@conan_kudo:matrix.org>
17:37:12
I would prefer that actually
<@zbyszek:fedora.im>
17:37:18
WFM too. I actually have something to attend.
<@jistone:fedora.im>
17:37:38
don't we need to approve election questions?
<@conan_kudo:matrix.org>
17:37:49
yes we do
<@nirik:matrix.scrye.com>
17:37:56
oh yeah
<@conan_kudo:matrix.org>
17:37:58
but we can do that async with a fast track
<@conan_kudo:matrix.org>
17:38:07
if they're not ready to approve right now
<@nirik:matrix.scrye.com>
17:38:57
yep
<@sgallagh:fedora.im>
17:39:34
!info 2024-11-26 meeting is CANCELED due to lack of expected quorum
<@sgallagh:fedora.im>
17:39:43
So who wants Dec. 3?
<@sgallagh:fedora.im>
17:40:24
!info F42/F43 Election questions will be reviewed and approved async using the FastTrack process
<@salimma:fedora.im>
17:40:25
not me (work travel) but I can probably do Dec 10
<@zbyszek:fedora.im>
17:40:42
I can take Dec. 3.
<@sgallagh:fedora.im>
17:41:19
Thanks
<@sgallagh:fedora.im>
17:41:29
!action zbyszek to chair 2024-12-03 meeting
<@sgallagh:fedora.im>
17:41:43
!topic Open Floor
<@sgallagh:fedora.im>
17:41:50
nirik: You had two items here?
<@nirik:matrix.scrye.com>
17:41:54
ok, fun thing number one:
<@nirik:matrix.scrye.com>
17:42:00
https://pagure.io/fedora-infrastructure/issue/12167
<@nirik:matrix.scrye.com>
17:42:04
!ticket 12167
<@zodbot:fedora.im>
17:42:05
● **Last Updated:** 2 months ago
<@zodbot:fedora.im>
17:42:05
**fedora-infrastructure #12167** (https://pagure.io/fedora-infrastructure/issue/12167):**bugzilla components have a 64 char limit**
<@zodbot:fedora.im>
17:42:05
● **Opened:** 2 months ago by kevin
<@zodbot:fedora.im>
17:42:05
● **Assignee:** Not Assigned
<@zodbot:fedora.im>
17:42:05
<@nirik:matrix.scrye.com>
17:42:14
turns out bugzilla has a 64 char limit on components.
<@nirik:matrix.scrye.com>
17:42:22
and we now have a component > 64 chars
<@nirik:matrix.scrye.com>
17:42:42
should we forbid that? how do we make it shorter?
<@salimma:fedora.im>
17:43:03
there's no mechanism for having package name be different from component name right now, is there?
<@nirik:matrix.scrye.com>
17:43:10
I did ask bugzilla admins, and it's not configrable, it's in the db schema... so, it's still possible to change, but will be a big thing and not until next year sometime
<@sgallagh:fedora.im>
17:43:29
nirik: Golang package?
<@nirik:matrix.scrye.com>
17:43:43
I think a lot of things assume name -> bz component name...
<@zbyszek:fedora.im>
17:43:45
Such long package names are annoying to users too.
<@nirik:matrix.scrye.com>
17:44:13
golang-github-azure-sdk-resourcemanager-resourcegraph-armresourcegraph
<@zbyszek:fedora.im>
17:44:14
I'd just say in the Packaging Guidelines to shorten the name to fit the limit.
<@nirik:matrix.scrye.com>
17:44:15
is the package
<@zbyszek:fedora.im>
17:44:31
We already forbid various characters in names, so upstream project names occasionally need to be mangled anyway.
<@nirik:matrix.scrye.com>
17:45:11
ok, so should we make this a fesco policy? or should I go to FPC to add it to packaging guidelines? or ?
<@sgallagh:fedora.im>
17:45:19
Shortening that as a package could be problematic with the way that golang packaging generates dependencies
<@salimma:fedora.im>
17:45:33
it wouldn't, I think
<@salimma:fedora.im>
17:45:43
the golang(...) dependencies are based on the goipath
<@jistone:fedora.im>
17:45:47
it may take some extra effort over normal golang packages, but surely it can still provide a longer name?
<@nirik:matrix.scrye.com>
17:46:02
We don't have to decide anything today, but I wanted to bring it up so we could fix it and I could stop getting an email every day about it. ;)
<@sgallagh:fedora.im>
17:46:03
Josh Stone: Yeah, true
<@sgallagh:fedora.im>
17:46:25
Josh Stone: For a second there, I read that as sarcasm: "surely they could make that name longer if they tried"
<@zbyszek:fedora.im>
17:46:33
Let's approve the limit and let golang packagers figure out the details of how to deal with the mapping.
<@salimma:fedora.im>
17:46:36
best check with the golang SIG folks (the core ones that wrote the tooling, not me) but I *think* the name does not matter
<@salimma:fedora.im>
17:46:36
<@salimma:fedora.im>
17:46:36
we have had compat epel-only golang packages where the name is different and they are drop in replacements
<@sgallagh:fedora.im>
17:46:51
nirik: My recommendation is that we NOT try to adapt Bugzilla for it, since that's also going away with the new forge
<@nirik:matrix.scrye.com>
17:46:59
right
<@nirik:matrix.scrye.com>
17:48:02
So, whats the path forward. I can just file a bug on... oh wait, no, I can't file a bug on the package. ;)
<@salimma:fedora.im>
17:48:17
maybe file a bug against go2rpm
<@salimma:fedora.im>
17:48:27
since that's what is used to generate new packages
<@decathorpe:fedora.im>
17:48:30
you can override the `Name` for golang packages, the heuristics for shortening names in go2rpm are not set in stone (and not always good)
<@salimma:fedora.im>
17:48:57
make it spit out a warning if the name is > 64 and tell the packager to use their judgement for shortening it
<@salimma:fedora.im>
17:49:13
we need to move to nirik's fun issue number 2 soon :)
<@sgallagh:fedora.im>
17:49:20
nirik: I think we respond on the package review BZ that the name can't be used for technical reasons and that they need to shorten it.
<@decathorpe:fedora.im>
17:49:22
or, ideally, add a hook into fedora-review or fedora-scm-requests ...
<@nirik:matrix.scrye.com>
17:49:42
ok, I'll try and move things forward in all those places.
<@nirik:matrix.scrye.com>
17:49:52
this one package needs renamed too.
<@nirik:matrix.scrye.com>
17:51:27
ok, second fun item: retirements. We have moved from an old rawhide script to a toddler process to handle things and there's some implementation details: it looks at if a release is frozen and doesn't do retirements then. That will affect beta freeze while it's happening, but then once if lifts it will block all the retired packages. For final it means no retirements after freeze starts.
<@nirik:matrix.scrye.com>
17:52:03
I guess I should write up a more detailed thing here for list/discussion. We may want to re-figure or change retirements...
<@sgallagh:fedora.im>
17:52:12
!info nirik raised an issue with a package whose name exceeded the maximum size of components for Bugzilla. FESCo recommends that the package be renamed in Fedora, citing technical limitations. We should also document this and, ideally, update fedora-package-review.
<@salimma:fedora.im>
17:52:56
Stephen Gallagher: don't forget updating fedora-scm-requests
<@zbyszek:fedora.im>
17:53:17
Hmm, why is this a problem? Package retirements are not allowed after Final Freeze, iiuc.
<@salimma:fedora.im>
17:53:17
fedora-review is in a bit of a sad state ... *cough* tech debt
<@sgallagh:fedora.im>
17:53:27
!info Addendum: also modify fedora-scm-requests to detect this.
<@decathorpe:fedora.im>
17:53:53
`fedpkg request-repo <long name here>` failing would be a bit too late though, at that point the package was already approved.
<@decathorpe:fedora.im>
17:54:13
it would be better if this were caught during the review. maybe the review bot can do it?
<@zbyszek:fedora.im>
17:54:32
fedora-package-review seems like the most appropriate place. It shouldnt' be hard to add a check .
<@conan_kudo:matrix.org>
17:54:43
when did this happen?
<@nirik:matrix.scrye.com>
17:54:59
zbyszek: our current policy says: "But this is only possible for development releases such as Branched or Rawhide."and "Packages will be retired in Fedora Rawhide and Branched (until the Beta Freeze)."
<@zbyszek:fedora.im>
17:55:31
Hmm, where is this from?
<@nirik:matrix.scrye.com>
17:55:44
https://docs.fedoraproject.org/en-US/fesco/Policy_for_orphan_and_retired_packages/
<@sgallagh:fedora.im>
17:55:45
Conan Kudo: It's been part of the plan for multiple years now. The selected git forge's issue tracker will replace Bugzilla for packaging tickets.
<@conan_kudo:matrix.org>
17:55:55
...
<@zbyszek:fedora.im>
17:55:59
https://docs.fedoraproject.org/en-US/package-maintainers/Package_Retirement_Process/#what_can_be_retired disagrees :(
<@nirik:matrix.scrye.com>
17:56:09
see! fun issues!
<@conan_kudo:matrix.org>
17:56:18
it definitely _was not_ part of any plan publicly communicated that I've seen
<@salimma:fedora.im>
17:56:19
yeah but as a failsafe... not everyone runs fedora-review, it's better to have it in both places I think
<@conan_kudo:matrix.org>
17:56:22
but whatever
<@zbyszek:fedora.im>
17:56:50
I think we want to allow retirement between Beta Freeze and Final Freeze.
<@nirik:matrix.scrye.com>
17:56:53
I think it was proposed/discussed, but there's no approved plan (yet?)
<@decathorpe:fedora.im>
17:56:54
it wasn't explicitly, but *implicitly* there was a lot of *wink wink*
<@conan_kudo:matrix.org>
17:57:02
that's the worst
<@humaton:fedora.im>
17:57:21
while working in releng I wa susing https://docs.fedoraproject.org/en-US/package-maintainers/Package_Retirement_Process/
<@zbyszek:fedora.im>
17:57:22
So I'd propose to change Policy for orphan… document to match Package Retirement Process document.
<@nirik:matrix.scrye.com>
17:57:51
sure, I can submit a pr...
<@decathorpe:fedora.im>
17:58:27
this one looks quite outdated - https://docs.fedoraproject.org/en-US/fesco/Policy_for_orphan_and_retired_packages/
<@humaton:fedora.im>
17:58:36
Conan Kudo: https://pagure.io/Fedora-Council/tickets/issue/473#comment-904965
<@farchord:fedora.im>
17:58:44
Wait, bugzilla is going away? I thought only pagure-parts were changing
<@sgallagh:fedora.im>
17:58:50
Conan Kudo: https://discussion.fedoraproject.org/t/arc-git-forge-investigation/114018 See points 2 and 3, specifically
<@salimma:fedora.im>
17:59:04
the council decision on the forge might be happening this week
<@farchord:fedora.im>
17:59:33
Ahh..... interestibg
<@humaton:fedora.im>
17:59:34
it was part of the assignment we got from council.
<@zbyszek:fedora.im>
17:59:45
OK, nirik, is adjusting the docs enough to resolve the original question about the toddler process?
<@nirik:matrix.scrye.com>
17:59:51
Steve Cossette [Farchord]: there's no retirement date I know of, but... rhel is no longer using it. So, I can't think that it will continue forever.
<@conan_kudo:matrix.org>
18:00:11
and switching to a more vanilla bugzilla wasn't on the table?
<@nirik:matrix.scrye.com>
18:00:21
perhaps. I could note the beta freeze thing (and notify people on list/discussion about it) and perhaps thats enough
<@conan_kudo:matrix.org>
18:00:24
that seems like a way more natural thing to explore too
<@conan_kudo:matrix.org>
18:00:40
what kind of process doesn't include options around actually having a proper bug tracker?
<@decathorpe:fedora.im>
18:00:47
so the actual difference between toddler behaviour and documented thing (https://docs.fedoraproject.org/en-US/package-maintainers/Package_Retirement_Process/) is that package retirements are "on hold" during beta freeze?
<@farchord:fedora.im>
18:01:07
It... depends. If the gitforge solution has a suitable bug tracker, then... might as well use that
<@nirik:matrix.scrye.com>
18:01:09
yes, and retirements do happen between beta and final freezes
<@decathorpe:fedora.im>
18:01:10
a lot of the requirements that were gathered included bug tracker stuff, IIRC.
<@decathorpe:fedora.im>
18:01:34
ok, in that case, the change is minor, and I think it would be fine to just document it ...
<@salimma:fedora.im>
18:01:49
yeah. I'm going to attend just in case, because one requirement that's key to me is being able to have bugs depend on each other
<@salimma:fedora.im>
18:02:01
which apparently gitlab can't do
<@conan_kudo:matrix.org>
18:02:28
the reason they are split is because bugs.kde.org provides a global index of bugs with persistent ID URLs
<@conan_kudo:matrix.org>
18:02:44
even if a bug is shifted around, the ID is static, which means it's always accessible
<@sgallagh:fedora.im>
18:02:57
Michel Lind 🎩 UTC-6: Last I heard, forgejo is the lead option and it DOES support issue dependencies
<@conan_kudo:matrix.org>
18:03:17
forge trackers are designed around the idea you don't have those things and each project is self-contained rather than part of a greater whole
<@salimma:fedora.im>
18:03:22
that's what I hear too
<@conan_kudo:matrix.org>
18:03:25
it's a huge impedence mismatch
<@conan_kudo:matrix.org>
18:03:46
that's why we never used pagure issue tracking either
<@salimma:fedora.im>
18:03:47
but yeah dealing with misfiled bugs is the other issue that dedicated issue trackers are better at
<@humaton:fedora.im>
18:03:57
not true in most of them there are Organizations that allow you to group bugz in our case by releases
<@salimma:fedora.im>
18:04:03
anyway, that's something to tell Council and not us here :)
<@dcantrell:fedora.im>
18:04:15
yes, bring this to the council discussion please
<@farchord:fedora.im>
18:04:18
As long as the issue tracker in the git forge can be 100% disabled, that's fine
<@salimma:fedora.im>
18:04:21
also... are we out of time?
<@sgallagh:fedora.im>
18:04:32
nirik: Could you !info the retirement topic, please
<@zbyszek:fedora.im>
18:04:33
No, there is no time limit to this meeting.
<@salimma:fedora.im>
18:04:41
I mean if that replaces bugzilla. but anyway, off topic :)
<@sgallagh:fedora.im>
18:04:47
Michel Lind 🎩 UTC-6: We technically have two hours alotted, but try not to use it
<@salimma:fedora.im>
18:04:53
ah right, I forgot. and the calendar actually alloted 90 mins anyway
<@salimma:fedora.im>
18:04:56
s/90/120
<@nirik:matrix.scrye.com>
18:04:57
!info nirik to make a pr on policy around retirements and when they happen (and note that they happen between beta and final freezes)
<@zbyszek:fedora.im>
18:05:36
Next topic?
<@sgallagh:fedora.im>
18:05:55
Does anyone else have an Open Floor topic?
<@zbyszek:fedora.im>
18:06:00
.fesco 3189
<@zbyszek:fedora.im>
18:06:06
!fesco 3189
<@zodbot:fedora.im>
18:06:08
● **Last Updated:** 7 months ago
<@zodbot:fedora.im>
18:06:08
● **Opened:** 7 months ago by amoloney
<@zodbot:fedora.im>
18:06:08
● **Assignee:** mattia
<@zodbot:fedora.im>
18:06:08
**fesco #3189** (https://pagure.io/fesco/issue/3189):**Change: Change Compose Settings**
<@zodbot:fedora.im>
18:06:08
<@zodbot:fedora.im>
18:06:08
● **Closed: Accepted** 7 months ago by zbyszek
<@zbyszek:fedora.im>
18:06:14
!fesco 3289
<@zodbot:fedora.im>
18:06:15
● **Assignee:** Not Assigned
<@zodbot:fedora.im>
18:06:15
● **Last Updated:** 21 hours ago
<@zodbot:fedora.im>
18:06:15
● **Opened:** a day ago by adamwill
<@zodbot:fedora.im>
18:06:15
<@zodbot:fedora.im>
18:06:15
**fesco #3289** (https://pagure.io/fesco/issue/3289):**Reconsider updates policy karma requirements due to possibly unintentional changes from 2021**
<@zbyszek:fedora.im>
18:06:16
Sorry!
<@zbyszek:fedora.im>
18:06:22
Can we approve this now?
<@salimma:fedora.im>
18:06:38
yes, there should be enough +1s already
<@zbyszek:fedora.im>
18:06:45
It's +7.
<@zbyszek:fedora.im>
18:06:55
But we'd need to wait another 6 days if we don't do it now.
<@nirik:matrix.scrye.com>
18:07:12
I'm +1, just hadn't gotten around to adding to ticket
<@sgallagh:fedora.im>
18:07:26
Who is missing?
<@humaton:fedora.im>
18:07:29
oh yeah I thought I did that
<@humaton:fedora.im>
18:07:34
+1
<@dcantrell:fedora.im>
18:07:41
count me as +1
<@zbyszek:fedora.im>
18:07:47
jednorozec I see your vote.
<@humaton:fedora.im>
18:07:49
oh I did in the ticket nevermind
<@decathorpe:fedora.im>
18:07:56
(just for the record - does bodhi 8.2 support setting different thresholds in config?)
<@decathorpe:fedora.im>
18:08:17
not that we approve something that can't currently be implemented and would require more code changes
<@nirik:matrix.scrye.com>
18:08:36
I think it does... adamw likely knows, he reworked it
<@zbyszek:fedora.im>
18:08:38
OK, so that's +9 and according to the voting rules, the ticket is approved.
<@sgallagh:fedora.im>
18:09:00
zbyszek: Will you update the ticket?
<@adamwill:fedora.im>
18:09:13
yes, it does
<@zbyszek:fedora.im>
18:09:19
OK.
<@adamwill:fedora.im>
18:09:19
changing this is a one byte change in ansible
<@nirik:matrix.scrye.com>
18:09:36
Oh hey. I thought of one more fun thing? :) Don't you all love me for it?
<@sgallagh:fedora.im>
18:10:13
I can't help but wonder how many hours of fixes that one-byte change will require :-)
<@adamwill:fedora.im>
18:10:24
Stephen Gallagher: none. i'm very confident about this, fwiw
<@salimma:fedora.im>
18:10:39
"it is one byte, but the question is which byte"
<@salimma:fedora.im>
18:10:43
jk
<@sgallagh:fedora.im>
18:10:50
Right... time to go hide in my bunker.
<@nirik:matrix.scrye.com>
18:11:08
So, we had a f41 change to move to new createrepo_c and repodata. We implemented it for f41 and rawhide, but then we backed it out because rhel7 was still around. Then we did implement it for f41 updates.
<@zbyszek:fedora.im>
18:11:16
Stephen Gallagher I merged the PR now and commented in the ticket. Please mention it in the meeting summary.
<@adamwill:fedora.im>
18:11:27
oh, god, wait, i might be forgetting something
<@adamwill:fedora.im>
18:11:30
yeah i probably am
<@adamwill:fedora.im>
18:11:34
oh well, time for coding!
<@sgallagh:fedora.im>
18:11:37
zbyszek: Want to just !info it?
<@nirik:matrix.scrye.com>
18:11:38
so, I want to move rawhide to zstd again, but it's going to break rhel people again... how much announcement/where should I do for that?
<@adamwill:fedora.im>
18:11:55
yeah i lied. oh well
<@adamwill:fedora.im>
18:12:06
this will need a new bodhi release
<@sgallagh:fedora.im>
18:12:17
nirik: Define "break rhel people" please?
<@zbyszek:fedora.im>
18:12:20
!info #3289 Reconsider updates policy karma requirements due to possibly unintentional changes from 2021 was approved and the documentation change was merged.
<@adamwill:fedora.im>
18:12:21
i didn't actually separate critpath min karma and non-critpath min karma, cos they were the same/
<@decathorpe:fedora.im>
18:12:26
see, this is why I asked *before* approving it :D
<@salimma:fedora.im>
18:12:31
shouldn't RHEL 10+ support zstd?
<@adamwill:fedora.im>
18:12:37
Fabio Valentini 🤒: thanks for your service :)
<@decathorpe:fedora.im>
18:12:44
alas
<@nirik:matrix.scrye.com>
18:13:22
rhel8/9 createrepo_c has no zstd support. So, people who run koji's and try and use rawhide repos will have their newrepo jobs fail. Things like that.
<@nirik:matrix.scrye.com>
18:13:35
rhel10 probibly is fine
<@salimma:fedora.im>
18:13:48
oh right
<@salimma:fedora.im>
18:13:54
this could be bad for mirrors
<@nirik:matrix.scrye.com>
18:14:00
why?
<@salimma:fedora.im>
18:14:16
nevermind, duh, they just sync over the repo definitions and don't generate locally
<@nirik:matrix.scrye.com>
18:14:22
It's only if you are trying to mergerepo_c rawhide or something.
<@sgallagh:fedora.im>
18:14:42
nirik: Is there a ticket open asking for createrepo_c support to be backported to RHEL 9 and (ideally) 8?
<@nirik:matrix.scrye.com>
18:14:45
I'll at least start a discussion so people know it's coming...
<@nirik:matrix.scrye.com>
18:15:08
I'm not sure, I think there might have been? would have to look
<@nirik:matrix.scrye.com>
18:15:18
anyhow, thats all from me. I will stop now. ;)
<@sgallagh:fedora.im>
18:15:34
I feel like that's something we could probably talk them into doing, to be forward-compatible
<@sgallagh:fedora.im>
18:15:41
We did similar stuff with libmodulemd once upon a time
<@sgallagh:fedora.im>
18:16:42
!info Plans to switch Fedora repodata to use zstd compression are currently on hold due to incompatibility with RHEL 8 and 9 createrepo_c. Options are being explored.
<@dcantrell:fedora.im>
18:17:00
just request the support in rhel8 and rhel9
<@dcantrell:fedora.im>
18:17:05
hell, I'll take the bugs and do the changes
<@sgallagh:fedora.im>
18:17:09
dcantrell: Isn't that your team that maintains that?
<@dcantrell:fedora.im>
18:17:12
yes
<@nirik:matrix.scrye.com>
18:17:25
there might be some. I'd have to look.
<@nirik:matrix.scrye.com>
18:17:30
I guess you are telling me to look now. ;)
<@dcantrell:fedora.im>
18:17:59
if there are any open, tell me the BZ numbers or Jira numbers or Github issues or wherever the request exists since we have multiple firehoses of work these days
<@sgallagh:fedora.im>
18:18:18
But they're going to have to support it at some point, or Pulp will not be able to work with Fedora eventually
<@nirik:matrix.scrye.com>
18:18:45
s/eventually//
<@nirik:matrix.scrye.com>
18:18:50
f41 updates repo is zstd
<@sgallagh:fedora.im>
18:19:17
Right. That.
<@nirik:matrix.scrye.com>
18:19:30
jira hates me. I will look out of band
<@decathorpe:fedora.im>
18:21:20
zbyszek: maybe https://pagure.io/fesco/fesco-docs/pull-request/93 should have waited with merging until bodhi actually supports this?
<@decathorpe:fedora.im>
18:21:47
changing the policy now just re-creates the situation where the policy doesn't match bodhi behaviour :(
<@zbyszek:fedora.im>
18:22:06
I don't think so. IIUC, the implementation will be updated to match docs soon.
<@decathorpe:fedora.im>
18:22:22
no pressure, adamw ...
<@nirik:matrix.scrye.com>
18:22:24
https://issues.redhat.com/browse/RHEL-67689
<@zbyszek:fedora.im>
18:22:27
We have to do some steps earlier than other steps.
<@decathorpe:fedora.im>
18:22:48
approve -> implement -> document would be my intiution, but WDIK
<@sgallagh:fedora.im>
18:23:16
OK, do we have any more topics, or are we good here for today?
<@sgallagh:fedora.im>
18:24:37
Alright, thank you for coming, folks.
<@sgallagh:fedora.im>
18:24:42
!endmeeting