#meeting-1:fedoraproject.org log

<@james:fedora.im>

17:00:46

!startmeeting fpc

<@meetbot:fedora.im>

17:00:48

Meeting started at 2024-12-05 17:00:46 UTC

<@meetbot:fedora.im>

17:00:48

The Meeting name is 'fpc'

<@james:fedora.im>

17:00:50

!topic Roll Call

<@limb:fedora.im>

17:00:56

!hi

<@zodbot:fedora.im>

17:00:58

Gwyn Ciesla (limb) - she / her / hers

<@james:fedora.im>

17:01:24

!hi

<@zodbot:fedora.im>

17:01:25

James Antill (james)

<@salimma:fedora.im>

17:02:17

!hi

<@zodbot:fedora.im>

17:02:18

Michel Lind (salimma) - he / him / his

<@carlwgeorge:matrix.org>

17:04:40

!hi

<@zodbot:fedora.im>

17:04:42

Carl George (carlwgeorge) - he / him / his

<@tibbs:fedora.im>

17:04:44

Hey.

<@james:fedora.im>

17:06:49

Okay, that's five of us...

<@james:fedora.im>

17:07:12

!topic FPC 1415 - https://pagure.io/packaging-committee/issue/1415

<@james:fedora.im>

17:08:56

Saw this go past in infra. places, and figured we should have something about it ... new forge might help some things, but they'll still be limits somewhere.

<@decathorpe:fedora.im>

17:09:07

hey o/

<@decathorpe:fedora.im>

17:09:07

sorry, I won't be able to participate (much) today - have a pretty bad headache

<@james:fedora.im>

17:10:23

No problem, I know a bunch of people who've got colds in the last week.

<@tibbs:fedora.im>

17:10:42

The question is whether there just needs to be a note somewhere, or if we need to outright ban >64 character package names.

<@decathorpe:fedora.im>

17:10:46

yeah, basically everyone I know has been sick in the past 2-3 weeks

<@limb:fedora.im>

17:10:50

There's definately something going around.

<@tibbs:fedora.im>

17:11:17

I guess we can just go off of what releng suggests that we do. But it's dumb to change it and then change it back.

<@limb:fedora.im>

17:11:27

I feel like we should specify that limit, and increase later if possible.

<@decathorpe:fedora.im>

17:11:27

as I mentioned on the ticket - I think it's premature (or too late?) to make a rule now because of bugzilla

<@limb:fedora.im>

17:12:01

Maybe less a rule, and more a note that >64 won't work.

<@james:fedora.im>

17:12:19

Fabio Valentini: How fast do you think the transition from bugzilla will be?

<@decathorpe:fedora.im>

17:12:19

it might be a good idea so define a "soft limit" and warn packagers about really long names *somewhere* but I don't think bugzilla (which is on its way to a farm upstate) should dictate a hard limit (it's too late for that)

<@limb:fedora.im>

17:12:46

To what will we be moving from bugzilla?

<@decathorpe:fedora.im>

17:12:47

James Antill: there's apparently been a "Unified Decision" to move to Forgejo ;)

<@salimma:fedora.im>

17:13:12

Agreed, this should be an advisory thing - by the way we know this will break - and we shouldn't lock that design for eternity

<@conan_kudo:matrix.org>

17:14:01

!hi

<@zodbot:fedora.im>

17:14:03

Neal Gompa (ngompa) - he / him / his

<@james:fedora.im>

17:14:45

Fabio Valentini: Yeh, but I'm still not clear on the timeline ... might be 2025, but I wouldn't bet a lot on it.

<@james:fedora.im>

17:14:53

Esp. on the bugzilla part.

<@decathorpe:fedora.im>

17:15:22

wibbly-wobbly timey-wimey

<@james:fedora.im>

17:15:42

I believe that's Conan Kudo 's line ;)

<@conan_kudo:matrix.org>

17:15:48

<@salimma:fedora.im>

17:15:50

As long as we don't get into a mailing list vs discourse split

<@conan_kudo:matrix.org>

17:15:53

tbh, I don't expect particular progress on that until 2026 at least

<@decathorpe:fedora.im>

17:16:12

hey, I know my Doctor Who too

<@salimma:fedora.im>

17:16:33

Who?

<@salimma:fedora.im>

17:16:35

Kidding

<@carlwgeorge:matrix.org>

17:16:45

From what I hear there qa team has concerns around forgejo issue tracking, and one suggestion was we disconnect moving dist-git from moving bugs (i.e. keep bugs in bugzilla longer)

<@conan_kudo:matrix.org>

17:16:48

I was going to be so disappointed if you didn't know :P

<@conan_kudo:matrix.org>

17:17:18

I share those concerns as well

<@decathorpe:fedora.im>

17:17:22

Proposal: Add a warning for source package name >64 characters to the Package Review guidelines and / or fedora-review.

<@tibbs:fedora.im>

17:17:40

I have no doubt we will spend a huge amount of effort to move forges an still not get what we have with pagure.

<@tibbs:fedora.im>

17:18:33

+1 to adding some note somewhere.

<@decathorpe:fedora.im>

17:19:32

maybe *bonk*ing go2rpm too, it seems to be the main culprit in making really long package names.

<@carlwgeorge:matrix.org>

17:19:35

I like this because a limit of the tooling would be noted in other tooling, versus policy

<@james:fedora.im>

17:20:37

To be fair, the original infra. ticket (https://pagure.io/fedora-infrastructure/issue/12167) has notes to change go2rpm and fedora-review ... great minds think alike ;)

<@james:fedora.im>

17:20:55

I just figured we should have something in policy for when someone asks/complains

<@tibbs:fedora.im>

17:21:08

You will find that a whole bunch of the packaging guidelines are exactly related to limits of the tooling (specifically RPM).

<@decathorpe:fedora.im>

17:21:14

James Antill: great thinkers plagiarize even better thinkers

<@james:fedora.im>

17:23:05

Anyone want to volunteer to do a PR?

<@tibbs:fedora.im>

17:23:43

I am about to go out of town for a week so I can't volunteer this time.

<@salimma:fedora.im>

17:24:06

Pretty sure the same people here discussed it anyway 😅. Self plagiarism?

<@james:fedora.im>

17:24:42

The best kind :)

<@salimma:fedora.im>

17:24:46

I have a bunch of writeup I need to do for the next few days, so while I'm doing that I might as well do this too

<@james:fedora.im>

17:26:55

!topic Open Floor

<@james:fedora.im>

17:27:39

Next week likely to be the last meeting of the year.

<@tibbs:fedora.im>

17:27:58

Was there a PHP-related request that we needed to addres?

<@tibbs:fedora.im>

17:28:32

The PR list got a good bit longer recently with the license change stuff which we should also address.

<@james:fedora.im>

17:28:44

Technically I think I'll be around on the 19th, but I know a lot of people are already on holiday by then ... and not much is likely to need us.

<@james:fedora.im>

17:29:09

Ahh, probably need to look through the ones that aren't tagged for meeting.

<@tibbs:fedora.im>

17:29:18

Once I get back next week I'll be fully available.

<@james:fedora.im>

17:29:39

Ahh, yeh, msuchy opened a bunch.

<@tibbs:fedora.im>

17:29:41

I really wish there weren't 97 open issues.

<@tibbs:fedora.im>

17:30:03

The PHP thing was https://pagure.io/packaging-committee/issue/1401

<@tibbs:fedora.im>

17:30:23

I guess the issue here is that there's no PR.

<@carlwgeorge:matrix.org>

17:32:06

Yeah I remember that one being odd as like a description of a diff instead of an actual pr

<@tibbs:fedora.im>

17:34:07

All we can really do now besides ask for a PR is to comment on the things we can see in that ticket. I'm not going to have any real opinion on PHP things.

<@james:fedora.im>

17:34:08

https://docs.fedoraproject.org/en-US/packaging-guidelines/#renaming-or-replacing-existing-packages

<@james:fedora.im>

17:34:08

Also ... what does everyone think of: https://pagure.io/packaging-committee/pull-request/1412 ... adds text to the end of

<@james:fedora.im>

17:34:45

My first reaction was "WTF, of course we tell people about that" ... but I think I've just written text in emails or other places.

<@tibbs:fedora.im>

17:34:49

Regarding 1412 I think it's surprisingly long for what I'd hope would be a simple thing.

<@james:fedora.im>

17:34:56

But maybe it's somewhere else?

<@conan_kudo:matrix.org>

17:36:37

it's documented in rpm I believe?

<@carlwgeorge:matrix.org>

17:36:59

The obsoletes self thing confuses me, that is implicit when the name is the same

<@salimma:fedora.im>

17:37:10

I generally prefer explicit over implicit - so to me the length is fine, but yeah what Carl said

<@salimma:fedora.im>

17:37:31

maybe there needs to be an explicit example for the self-obsolete case because that confuses me too

<@carlwgeorge:matrix.org>

17:38:35

In the example given I would use requires or recommends in bar to pull it in, and use conflicts to avoid mismatched version file conflicts

<@conan_kudo:matrix.org>

17:38:49

you need the self obsoletes so that it gets added to the transaction

<@conan_kudo:matrix.org>

17:39:02

if you explicitly declare obsoletes, it removes the normal upgrade candidate from the transaction

<@conan_kudo:matrix.org>

17:39:18

so you have to do both to have both come in with the same transaction

<@james:fedora.im>

17:41:37

I think there's two general cases: foo becomes foo-core and foo-extra where foo-extra requires foo-core ... then foo is just renamed to foo-extra and the requires works out.

<@james:fedora.im>

17:41:37

That's if there's no requires, right?

<@tibbs:fedora.im>

17:41:41

And here we have another example of policy dealing with some quirk of the tooling.

<@conan_kudo:matrix.org>

17:41:51

yes, that's if there's no requires

<@conan_kudo:matrix.org>

17:41:58

here's the old fedora page about it: https://fedoraproject.org/wiki/Upgrade_paths_%E2%80%94_renaming_or_splitting_packages

<@james:fedora.im>

17:42:20

Ahh, we did have text about it at one point.

<@conan_kudo:matrix.org>

17:42:45

the opensuse page unfortunately is more detailed but less relevant because they implemented hints for making splitting smarter

<@conan_kudo:matrix.org>

17:42:55

https://en.opensuse.org/openSUSE:Package_dependencies#Splitting_and_Merging

<@conan_kudo:matrix.org>

17:43:33

the documentation around this stuff is very old

<@conan_kudo:matrix.org>

17:46:20

new documentation needs to be based on what libsolv behavior we configure

<@conan_kudo:matrix.org>

17:46:33

since libsolv supports multiple obsoletes strategies

<@conan_kudo:matrix.org>

17:47:16

but the basics are "every upgrade candidate you want to be included in a transaction needs the obsoletes line"

<@conan_kudo:matrix.org>

17:47:46

some folks like to use Conflicts+Obsoletes to represent something similar to Debian's Breaks+Replaces

<@conan_kudo:matrix.org>

17:48:04

(essentially, a one-way path)

<@conan_kudo:matrix.org>

17:48:51

(that's not what Breaks does: Breaks ensures that the package is uninstalled before the new package is installed)

<@conan_kudo:matrix.org>

17:49:12

Conflicts happens to create the same conditions as a side-effect

<@james:fedora.im>

17:55:33

I merged the License PRs ... and added a comment on the PHP issue about opening a PR.

<@james:fedora.im>

17:55:43

Anything else?

<@tibbs:fedora.im>

17:56:30

Not from me.

<@conan_kudo:matrix.org>

17:58:10

nothing from me

<@conan_kudo:matrix.org>

17:58:26

I spent almost a decade working in both ecosystems for work :)

<@conan_kudo:matrix.org>

17:58:38

and I maintain/develop debbuild

<@conan_kudo:matrix.org>

17:58:48

so I know how both rpm and dpkg work fairly well

<@james:fedora.im>

17:59:53

Jason ティビツ: Enjoy your week out of town.

<@james:fedora.im>

17:59:56

!endmeeting