<@zbyszek:fedora.im>
17:00:02
!startmeeting FESCO (2025-03-18)
<@meetbot:fedora.im>
17:00:05
Meeting started at 2025-03-18 17:00:02 UTC
<@meetbot:fedora.im>
17:00:05
The Meeting name is 'FESCO (2025-03-18)'
<@fale:fale.io>
17:00:07
!hi
<@zbyszek:fedora.im>
17:00:09
!meetingname fesco
<@zbyszek:fedora.im>
17:00:09
!topic Init Process
<@zbyszek:fedora.im>
17:00:09
Chairs: @conan_kudo:matrix.org, @ngompa:fedora.im, @nirik:matrix.scrye.com, @humaton:fedora.im, @zbyszek:fedora.im, @sgallagh:fedora.im, @fale:fale.io, @dcantrell:fedora.im, @decathorpe:fedora.im, @salimma:fedora.im
<@meetbot:fedora.im>
17:00:10
The Meeting Name is now fesco
<@conan_kudo:matrix.org>
17:00:24
!hi
<@zbyszek:fedora.im>
17:00:25
!hi
<@zodbot:fedora.im>
17:00:27
Neal Gompa (ngompa) - he / him / his
<@zodbot:fedora.im>
17:00:28
Zbigniew Jędrzejewski-Szmek (zbyszek)
<@zodbot:fedora.im>
17:00:30
Fabio Alessandro Locati (fale) - he / him / his
<@nirik:matrix.scrye.com>
17:01:21
morning
<@zbyszek:fedora.im>
17:01:35
We have 4/9 now.
<@humaton:fedora.im>
17:01:59
!hi
<@zodbot:fedora.im>
17:02:00
Tomáš Hrčka (humaton) - he / him / his
<@decathorpe:fedora.im>
17:02:15
!hi
<@zodbot:fedora.im>
17:02:16
Fabio Valentini (decathorpe) - he / him / his
<@zbyszek:fedora.im>
17:02:18
Let's start then.
<@zbyszek:fedora.im>
17:02:23
!topic #3364 F42 Incomplete Changes Report
<@zbyszek:fedora.im>
17:02:23
!fesco 3364
<@zbyszek:fedora.im>
17:02:46
!fesco 3364
<@zodbot:fedora.im>
17:02:47
● **Assignee:** Not Assigned
<@zodbot:fedora.im>
17:02:47
**fesco #3364** (https://pagure.io/fesco/issue/3364):**F42 Incomplete Changes Report**
<@zodbot:fedora.im>
17:02:47
<@zodbot:fedora.im>
17:02:47
● **Opened:** 4 weeks ago by amoloney
<@zodbot:fedora.im>
17:02:47
● **Last Updated:** 15 minutes ago
<@zbyszek:fedora.im>
17:02:53
:(
<@salimma:fedora.im>
17:02:59
!hi
<@zodbot:fedora.im>
17:03:15
Michel Lind (salimma) - he / him / his
<@zbyszek:fedora.im>
17:03:15
!topic #3364 F42 Incomplete Changes Report - Anaconda as native Wayland application
<@zbyszek:fedora.im>
17:03:15
I'll go one by one.
<@zbyszek:fedora.im>
17:03:54
jednorozec: I think we had an action item for you to check the status here…
<@humaton:fedora.im>
17:04:02
yup
<@humaton:fedora.im>
17:04:11
I am in anmother meeting but I have an update
<@zbyszek:fedora.im>
17:04:20
> This is already done, there is smaller subset of change which is currently last piece in BODHI but I wouldn't say it is blocking for the change. Everything else is merged months ago. https://bodhi.fedoraproject.org/updates/FEDORA-2025-a3429bf4d0
<@nirik:matrix.scrye.com>
17:04:22
there were updates on the ticket this morning.
<@zbyszek:fedora.im>
17:04:30
This is what Jiri wrote^
<@humaton:fedora.im>
17:05:07
yeah I talked to him this morning
<@nirik:matrix.scrye.com>
17:05:11
So, I think the only one left is the noaudit rules
<@nirik:matrix.scrye.com>
17:05:24
So, I think the only one left is the dontaudit rules
<@zbyszek:fedora.im>
17:05:49
OK, so are we in agreement that the two Anaconda items are done?
<@zbyszek:fedora.im>
17:06:12
!info The Change Owner reports that this is done.
<@zbyszek:fedora.im>
17:06:15
!link https://bodhi.fedoraproject.org/updates/FEDORA-2025-a3429bf4d0
<@nirik:matrix.scrye.com>
17:06:18
I think so...
<@zbyszek:fedora.im>
17:06:30
!topic #3364 F42 Incomplete Changes Report - Anaconda WebUI Partitioning
<@zbyszek:fedora.im>
17:06:40
!info The Change Owner reports that this is done.
<@zbyszek:fedora.im>
17:06:47
!topic #3364 F42 Incomplete Changes Report - Koji uses Red Hat Image Builder locally
<@zbyszek:fedora.im>
17:06:47
!info This Change has been retargeted to F43.
<@zbyszek:fedora.im>
17:06:57
!topic #3364 F42 Incomplete Changes Report - Reduce the amount of "dontaudit" rules pertaining to unlabeled_t
<@zbyszek:fedora.im>
17:07:00
This is the last one.
<@nirik:matrix.scrye.com>
17:09:21
Absent any info we should probibly just move this back to incomplete and ask them to resubmit when they are done? or should we give them more time?
<@zbyszek:fedora.im>
17:09:24
No reply in 25 days to my NEEDINFO :(
<@zbyszek:fedora.im>
17:10:37
Oh, this was already retargeted once…
<@salimma:fedora.im>
17:10:45
yeah
<@salimma:fedora.im>
17:10:48
I'd say mark this incomplete
<@zbyszek:fedora.im>
17:11:04
OK, let's make this formal.
<@salimma:fedora.im>
17:11:47
having been on the receiving end of this change for some packages - the proposer seems to expect the package maintainers to do all the work of verifying if the suggested changes makes sense and to upstream them. Which... no.
<@fale:fale.io>
17:11:55
+1
<@salimma:fedora.im>
17:12:01
if they're not driving this anymore, let's just vote on dropping it for now
<@zbyszek:fedora.im>
17:12:03
Proposal: Reduce the amount of "dontaudit" rules pertaining to unlabeled_t Change is marked as Incomplete. Owners can resubmit it for F43 or later if appropriate.
<@nirik:matrix.scrye.com>
17:12:10
+1
<@salimma:fedora.im>
17:12:14
+1
<@humaton:fedora.im>
17:12:22
+1
<@zbyszek:fedora.im>
17:13:07
Michel Lind UTC-6: can you link some ticket where this was discussed? I never saw anything about this.
<@zbyszek:fedora.im>
17:13:09
!agreed APPROVED (+5, 0, 0)
<@zbyszek:fedora.im>
17:13:32
I guess somebody didn't vote… Fabio Valentini ?
<@zbyszek:fedora.im>
17:14:44
!action Aoife Moloney to reassign the issue back to Incomplete
<@salimma:fedora.im>
17:14:51
If I remember the package where this happens... one sec. I think I remember one
<@decathorpe:fedora.im>
17:14:51
+1
<@decathorpe:fedora.im>
17:14:53
sorry :)
<@salimma:fedora.im>
17:15:11
!link https://src.fedoraproject.org/rpms/trousers/pull-request/10
<@conan_kudo:matrix.org>
17:15:13
+1
<@salimma:fedora.im>
17:15:33
zbyszek: "I am not familiar with this software and I don't know to test"
<@zbyszek:fedora.im>
17:15:58
!agreed APPROVED (+6, 0, 0)
<@zbyszek:fedora.im>
17:15:59
!fesco 3377
<@zbyszek:fedora.im>
17:15:59
!topic #3377 cmake 4.0 breaks compatibility, was updated without a Change Request
<@zbyszek:fedora.im>
17:15:59
There were some comments recently…
<@zbyszek:fedora.im>
17:17:18
I think we're in general in agreement that we want to undo the update.
<@humaton:fedora.im>
17:17:30
huh
<@zbyszek:fedora.im>
17:18:02
1. untag the update
<@zbyszek:fedora.im>
17:18:02
3. revert the commit, use fake Version: 4.0.0~really3.x
<@zbyszek:fedora.im>
17:18:02
I saw the following proposals:
<@zbyszek:fedora.im>
17:18:02
2. revert the commit, bump epoch
<@salimma:fedora.im>
17:18:05
from the devel discussion it seems the author of the commmit removing the settings also agrees to submit a CP and to revert
<@zbyszek:fedora.im>
17:18:19
I think untagging would be the nicest, if it works…
<@salimma:fedora.im>
17:18:23
I'd rather not do #2
<@salimma:fedora.im>
17:18:25
yeah
<@nirik:matrix.scrye.com>
17:18:27
well, does revert get us much?
<@nirik:matrix.scrye.com>
17:18:41
it's been in rawhide for almost a month now
<@salimma:fedora.im>
17:18:44
telling Rawhide users to 'distro-sync' cmake should be fine right?
<@zbyszek:fedora.im>
17:18:48
This was only in rawhide…
<@salimma:fedora.im>
17:18:56
at some point we'll get cmake 4.0 in Rawhide anyway
<@nirik:matrix.scrye.com>
17:19:12
does it break some people that already adjusted to 4.0?
<@conan_kudo:matrix.org>
17:19:17
yeah
<@conan_kudo:matrix.org>
17:19:23
it doesn't break anyone already
<@conan_kudo:matrix.org>
17:19:35
as long as you didn't update to minver 4.0 which nobody advised
<@decathorpe:fedora.im>
17:19:59
nirik: from what I can tell, there's an unknown amout of both silent and non-silent breakage with 4 and setting min_cmake_version to 3.5 unconditionally
<@nirik:matrix.scrye.com>
17:20:14
well, it breaks two groups then right? those that adjusted to 4.0 being there and depending on it or some feature, or those that were not yet adjusted and depended on the old one
<@nirik:matrix.scrye.com>
17:20:34
I would like to see the results of those mass prebuilds
<@conan_kudo:matrix.org>
17:20:40
there are no adaptations for 4.0 that don't work for 3.x
<@conan_kudo:matrix.org>
17:21:01
unless you set the cmake minimum version to 4.0 in cmakelists, everything is fine
<@music:fedora.im>
17:21:06
I’ve adapted quite a few packages to CMake 4.0 so far, and I haven’t encountered anything for which the most obvious fix was backwards-incompatible. I suppose it’s *possible* to do something backwards-incompatible.
<@music:fedora.im>
17:21:33
(Conan Kudo pointed out the obvious way to do so. I doubt anyone chose to do that.)
<@zbyszek:fedora.im>
17:22:35
cmake-4.0.0~rc2-1.fc43 was built 2025-02-26, and it was only in rawhide. This is just a short time and people don't generally develop only for rawhide.
<@nirik:matrix.scrye.com>
17:23:27
On one hand it seems correct to revert it and get a change. On the other it seems silly to revert it, get a change, approve the change and re-land it.
<@salimma:fedora.im>
17:24:03
we don't know how long it will take to get a change proposal for this approved though
<@fale:fale.io>
17:25:01
I think that if it is possible to get a cmake3 comp-package quickly enough it would make more sense not to revert for this time
<@conan_kudo:matrix.org>
17:25:24
and this change will need to detail remedies for packagers
<@conan_kudo:matrix.org>
17:25:43
we already know what those remedies are, the problem is there's no communication about it
<@conan_kudo:matrix.org>
17:26:14
and a cmake3 package will _still_ require people to adjust specfiles
<@zbyszek:fedora.im>
17:26:23
I think it's a bigger problem than just the version. The changes to "non-standard installation settings" seem unecessary.
<@conan_kudo:matrix.org>
17:26:47
yes, we don't have to drop those
<@conan_kudo:matrix.org>
17:26:52
and we probably shouldn't
<@conan_kudo:matrix.org>
17:27:02
we basically can't
<@music:fedora.im>
17:27:11
We also don’t yet know, even roughly, how many packages are impacted. For the packages I’m responsible for, it seemed to be roughly in the 25-35% range before `%cmake` started to set the `CMAKE_MINIMUM_REQUIRED` environment variable as a compatibility hack. Most of those only needed the declared minimum version adjusted. Some had deeper issues.
<@zbyszek:fedora.im>
17:27:14
I don't particularly want to adjust hundreds of packages to `BR:cmake3` and be stuck with the dependency on cmake3 for a long time.
<@conan_kudo:matrix.org>
17:27:31
we could add some macro tunables for the legacy flags to be able to turn them off
<@conan_kudo:matrix.org>
17:27:38
and then have a future change to flip it the other way
<@conan_kudo:matrix.org>
17:28:01
but deleting them whole-hog without communication with all the packagers is not really tenable
<@conan_kudo:matrix.org>
17:28:33
waiting for mpb is not tenable right now
<@zbyszek:fedora.im>
17:28:54
Proposal: We ask releng to untag cmake-4.0.x builds from rawhide and the maintainers to revert the update in dist-git and rebuild version 3.x. A Change proposal should be then filed and discussed to figure out how to do the transition in a smoother way.
<@conan_kudo:matrix.org>
17:28:56
rawhide is broken atm for lots of packages
<@decathorpe:fedora.im>
17:29:37
I think Cristian Le was running a MPB in copr right now, not sure if there are results yet.
<@conan_kudo:matrix.org>
17:29:44
zbyszek: is a rebuild required? the latest cmake 3.31.6 was build in rawhide
<@nirik:matrix.scrye.com>
17:29:46
and has been for a month? that seems surprising
<@conan_kudo:matrix.org>
17:29:55
before 4.0
<@nirik:matrix.scrye.com>
17:30:00
Fabio Valentini: there's 2... in the ticket. I don't know if they are done
<@conan_kudo:matrix.org>
17:30:22
nirik: the biggest stack (KDE) was only slightly impacted
<@conan_kudo:matrix.org>
17:30:26
it's everything else that's been a pain
<@zbyszek:fedora.im>
17:30:34
OK, let me rephrase.
<@conan_kudo:matrix.org>
17:30:40
right now the entire KF5 stack is borked
<@humaton:fedora.im>
17:30:41
the copr builds are pending
<@zbyszek:fedora.im>
17:30:48
Proposal: We ask releng to untag cmake-4.0.x builds from rawhide and the maintainers to revert the update in dist-git for now. A Change proposal should be then filed and discussed to figure out how to do the transition in a smoother way.
<@conan_kudo:matrix.org>
17:30:57
zbyszek: +1
<@nirik:matrix.scrye.com>
17:31:09
well, yeah, but this seems pretty rushed here... we are still trying to figure out the scope and stuff. Bunch of comments minutes before the meeting, it's been this way for a month... etc
<@zbyszek:fedora.im>
17:31:35
Hmm, many of the builds failed…
<@nirik:matrix.scrye.com>
17:31:40
if we are going to do that, I would also advocate a devel-announce post explaining that it was untagged...
<@conan_kudo:matrix.org>
17:31:58
yes of course
<@conan_kudo:matrix.org>
17:32:11
nirik: it was only fully bootstrapped 12 days ago
<@conan_kudo:matrix.org>
17:32:17
so the impact has only been felt for a week
<@humaton:fedora.im>
17:32:32
Bunch of them faild in both rebuilds...
<@music:fedora.im>
17:32:56
I actually think that change is worth *considering*, but I didn’t expect it to be summarily merged and shipped. It’s hard to discuss it when nobody has figured out the scope. If it were separately impact-checked, we would know how disruptive it was, or wasn’t, and could figure out what kind of proposal made sense, if any.
<@music:fedora.im>
17:33:29
But it is certainly not an imminently *necessary* change, in the sense that it’s not blocking anything.
<@zbyszek:fedora.im>
17:33:48
Yeah, I fully agree with what music said. We need to *discuss* this.
<@conan_kudo:matrix.org>
17:34:15
and I think there are softer ways to deal with dropping the nonstandard install setting flags
<@conan_kudo:matrix.org>
17:34:25
it needs some thought and discussion
<@zbyszek:fedora.im>
17:35:15
OK, I got one vote for the proposal. Please vote or submit a counterproposal.
<@nirik:matrix.scrye.com>
17:35:19
I guess I can be a weak +1 to the proposal... I am pretty sure someone(s) will pop up and complain about the revert, but such is the way it goes I guess.
<@decathorpe:fedora.im>
17:35:20
fwiw I'm +1 to zbyszek's proposal
<@humaton:fedora.im>
17:35:27
+1
<@conan_kudo:matrix.org>
17:35:53
that was really hard to scope and implement
<@zbyszek:fedora.im>
17:37:00
"music and nirik are typing …" — I'll wait
<@zbyszek:fedora.im>
17:37:54
Hmm, maybe it's stuck.
<@zbyszek:fedora.im>
17:38:10
I see that.
<@nirik:matrix.scrye.com>
17:38:27
I was typing, but in another room. ;)
<@zbyszek:fedora.im>
17:38:34
Oh, OK.
<@zbyszek:fedora.im>
17:38:44
!agreed (+5, 0, 0)
<@zbyszek:fedora.im>
17:39:24
!agreed We ask releng to untag cmake-4.0.x builds from rawhide and the maintainers to revert the update in dist-git for now. A Change proposal should be then filed and discussed to figure out how to do the transition in a smoother way.
<@zbyszek:fedora.im>
17:39:37
It's in two halves, but that should be OK.
<@zbyszek:fedora.im>
17:39:44
!topic Next week's chair
<@zbyszek:fedora.im>
17:39:44
\
<@zbyszek:fedora.im>
17:39:53
Vlntrs?
<@salimma:fedora.im>
17:40:29
I can take next week provided it does not run over time, I have a hard stop at 1:30
<@zbyszek:fedora.im>
17:40:58
Hmm, I have a house assoc. meeting next week. So I might not be here too.
<@nirik:matrix.scrye.com>
17:41:42
I had an item for open floor.
<@zbyszek:fedora.im>
17:41:59
!action Michel Lind UTC-6 will chair the next meeting
<@zbyszek:fedora.im>
17:42:02
!topic Open Floor
<@zbyszek:fedora.im>
17:42:08
Go, nirik.
<@nirik:matrix.scrye.com>
17:43:02
Oh, first, releng ticket for the cmake untagging: https://pagure.io/releng/issue/12639
<@zbyszek:fedora.im>
17:43:41
"just ended" — you're playing with our minds
<@nirik:matrix.scrye.com>
17:44:20
Then I wanted to ask if there's any reason we need to keep out container 'stuff' around... we have src.fedoraproject.org containers namespace, bodhi has fNC releases (with nothing in them) and bugzilla has components. But we don't build any of that anymore. Is there any objection to just removing it? or should I propose something for discussion about it? or ?
<@zbyszek:fedora.im>
17:44:21
"that will have just ended soon"?
<@nirik:matrix.scrye.com>
17:44:34
I'll just hop in my timemachine
<@zbyszek:fedora.im>
17:45:21
Hmm, there was some discussion on this topic in one of the bodhi updates…
<@decathorpe:fedora.im>
17:45:34
if none of that is used any longer, I'm fine with removing it, similarly to how "modules stuff" was dropped
<@nirik:matrix.scrye.com>
17:45:45
yeah, similar.
<@sgallagh:fedora.im>
17:45:45
Aargh, I got the time wrong. I'm here now
<@fale:fale.io>
17:46:09
I agree on dropping it if nothing of that is used anymore
<@sgallagh:fedora.im>
17:46:34
"modules stuff" just came up again recently
<@nirik:matrix.scrye.com>
17:46:59
It seems confusing to have it around
<@sgallagh:fedora.im>
17:47:03
DNF technically still supports it, which is why we still have PLATFORM_ID in os-release
<@nirik:matrix.scrye.com>
17:47:14
but we never formally decided anything around it
<@conan_kudo:matrix.org>
17:47:39
we should archive it somewhere, but I'm otherwise fine dropping it all like a hot potato
<@zbyszek:fedora.im>
17:47:45
https://fedoraproject.org/wiki/Changes/Drop_PLATFORM_ID
<@nirik:matrix.scrye.com>
17:48:25
sure, keep the existing data if we need it... but we don't need to keep making bodhi F(n)C releases or have it open for bugs in bugzilla, etc.
<@nirik:matrix.scrye.com>
17:48:45
I guess I can write up a list of things to do and see if anyone objects then
<@nirik:matrix.scrye.com>
17:49:14
thats all I had
<@zbyszek:fedora.im>
17:49:20
That sounds reasonable. Then you also have an URL to paste in the removal messages if things are tracked.
<@nirik:matrix.scrye.com>
17:49:38
Oh, as a FYI, the datacenter move looks to be moving to mid/late june now (from may)
<@fale:fale.io>
17:50:24
so after Flock :)
<@salimma:fedora.im>
17:50:35
so after devconf.cz? phew
<@fale:fale.io>
17:50:40
yeah, also after devconf.cz :)
<@sgallagh:fedora.im>
17:51:05
devconf.cz is after Flock, so yes to both :)
<@nirik:matrix.scrye.com>
17:51:20
yeah, after those
<@zbyszek:fedora.im>
17:52:02
OK, anything else for Open Floor?
<@zbyszek:fedora.im>
17:52:21
If not, I'll end in a minute.
<@zbyszek:fedora.im>
17:52:51
!endmeeting