#meeting-1:fedoraproject.org log

<@james:fedora.im>

16:00:23

!startmeeting fpc

<@meetbot:fedora.im>

16:00:24

Meeting started at 2024-04-18 16:00:23 UTC

<@meetbot:fedora.im>

16:00:24

The Meeting name is 'fpc'

<@james:fedora.im>

16:00:27

!topic Roll Call

<@james:fedora.im>

16:01:01

!hi

<@limb:fedora.im>

16:01:01

!hi

<@decathorpe:fedora.im>

16:01:01

!hi

<@zodbot:fedora.im>

16:01:03

James Antill (james)

<@zodbot:fedora.im>

16:01:05

Gwyn Ciesla (limb) - she / her / hers

<@zodbot:fedora.im>

16:01:06

Fabio Valentini (decathorpe) - he / him / his

<@zbyszek:fedora.im>

16:01:36

Hi, I'm also here to answer questions about https://pagure.io/packaging-committee/pull-request/1355.

<@conan_kudo:matrix.org>

16:01:46

!hi

<@zodbot:fedora.im>

16:03:26

Neal Gompa (ngompa) - he / him / his

<@tibbs:fedora.im>

16:06:07

Hello.

<@tibbs:fedora.im>

16:06:45

Sorry for being late; my client was stuck at startup again.

<@james:fedora.im>

16:08:41

No problem.

<@limb:fedora.im>

16:10:38

So I'm just now getting to reading this (sorry!) but I'm +1.

<@james:fedora.im>

16:11:15

!topic PR#1355 https://pagure.io/packaging-committee/pull-request/1355

<@james:fedora.im>

16:12:13

I thought the migration plan for that was all the %files be "correct" with what's the realpath on disk ... but packages would "Provide: /bin/foo" for binaries that were "well known" or whatever

<@conan_kudo:matrix.org>

16:12:24

I think I'm generally fine with it, but we may want to make a note of legacy locations MAY be provided by adding `Provides:` for them

<@conan_kudo:matrix.org>

16:12:48

to be honest, I thought this rule was already gone

<@decathorpe:fedora.im>

16:13:03

lgtm too, with Neal's addendum

<@conan_kudo:matrix.org>

16:13:04

it's basically untenable to maintain anyway

<@conan_kudo:matrix.org>

16:13:27

knowing what the "canonical" locations of things are supposed to be is tricky

<@limb:fedora.im>

16:14:36

I just use whatever the packagers worked out at the Council of Nicea but that's me.

<@salimma:fedora.im>

16:14:50

!hi

<@zodbot:fedora.im>

16:15:10

Michel Lind (salimma) - he / him / his

<@zbyszek:fedora.im>

16:16:06

Hmm, but what do you mean by the addendum? The rules already allow other paths to be specified, for compatibility, and packages do this quite a lot. It doesn't seem necessary or useful to repeat this here.

<@conan_kudo:matrix.org>

16:16:28

I think it contextually makes sense to remind them of this case given the sbin/bin merge

<@zbyszek:fedora.im>

16:18:00

The current text of the patch: + If other packages have dependencies on a different path, and it is not convenient to update them to the new path, packages **may** use a virtual `+Provides+` to list the alternate path. Isn't this enough?

<@james:fedora.im>

16:18:44

Sure

<@conan_kudo:matrix.org>

16:18:48

I would note that it makes sense as long as the path is actually reachable

<@conan_kudo:matrix.org>

16:18:57

(I know that should be obvious, but it sometimes isn't)

<@conan_kudo:matrix.org>

16:19:46

the path has to work to be a valid provides, and I actually don't recall us explicitly mentioning that anywhere

<@zbyszek:fedora.im>

16:20:29

Sure, I'll add that.

<@tibbs:fedora.im>

16:20:32

At some point you can always come up with some crazy thing that we don't explicitly mention.

<@tibbs:fedora.im>

16:20:53

If we mentioned all of them we'd have more of a tome than we have now.

<@conan_kudo:matrix.org>

16:21:10

lol

<@conan_kudo:matrix.org>

16:21:16

I think it just makes sense in this context to mention it

<@conan_kudo:matrix.org>

16:21:25

since we're talking about paths that only kind of exist

<@tibbs:fedora.im>

16:21:49

It's probably no more than making sure the word "valid" is in the right place.

<@conan_kudo:matrix.org>

16:22:18

anyway, to avoid more bikeshedding, with that bit it should be good to go

<@zbyszek:fedora.im>

16:22:24

"If other packages have dependencies on a different path that resolves to the same file, and it is not convenient to update them to the new path, packages **may** use a virtual `+Provides+` to list the alternate path." ?

<@james:fedora.im>

16:22:33

That seems like an autoQA type thing, where we could check that any virtual provides for a path actually has something at that path.

<@conan_kudo:matrix.org>

16:22:46

perfecto!

<@zbyszek:fedora.im>

16:24:07

I updated the pull request with that.

<@conan_kudo:matrix.org>

16:24:14

perfect

<@conan_kudo:matrix.org>

16:24:25

I'm good with it, and if everyone else is, I can merge it?

<@tibbs:fedora.im>

16:24:39

<@conan_kudo:matrix.org>

16:26:14

Anyone else object?

<@conan_kudo:matrix.org>

16:26:24

if not, I'll merge it now

<@conan_kudo:matrix.org>

16:27:36

Alright then...

<@conan_kudo:matrix.org>

16:28:02

!agreed The pull request is accepted and now merged.

<@zbyszek:fedora.im>

16:28:10

Thank you.

<@zbyszek:fedora.im>

16:28:12

I'll be back with a patch to nuke `%_sbindir` ;)

<@conan_kudo:matrix.org>

16:28:17

lol

<@conan_kudo:matrix.org>

16:29:34

so, anything else James Antill?

<@james:fedora.im>

16:30:19

Oh, y'all want me to be organised and everything ...

<@james:fedora.im>

16:31:07

!topic PR#1360 https://pagure.io/packaging-committee/pull-request/1360

<@james:fedora.im>

16:31:21

I think we need to wait on this, but not 100%

<@decathorpe:fedora.im>

16:32:31

waiting for the thing to be approved sounds good

<@decathorpe:fedora.im>

16:32:50

but it's great that the packaging guidelines patch is ready in advance 🙌🏼

<@limb:fedora.im>

16:32:51

Agreed. But once it's approved I think it makes a lot of sense.

<@james:fedora.im>

16:33:33

!topic PR#1357 https://pagure.io/packaging-committee/pull-request/1357

<@james:fedora.im>

16:33:40

I think this is a simple merge

<@conan_kudo:matrix.org>

16:33:57

yeah

<@conan_kudo:matrix.org>

16:34:19

that one is just more troubleshooting debuginfo stuff

<@limb:fedora.im>

16:34:35

I concur

<@conan_kudo:matrix.org>

16:35:03

I'll just merge it now

<@conan_kudo:matrix.org>

16:35:11

or nevermind

<@conan_kudo:matrix.org>

16:35:13

James beat me :D

<@james:fedora.im>

16:35:17

Okay, I merged it

<@james:fedora.im>

16:35:28

!topic PR#1358 https://pagure.io/packaging-committee/pull-request/1358

<@conan_kudo:matrix.org>

16:36:03

is that really how that works?

<@james:fedora.im>

16:36:13

I have no idea

<@tibbs:fedora.im>

16:36:13

I usually like to see Miro ack changes to the python guidelines.

<@decathorpe:fedora.im>

16:36:20

also ... no semantic line breaks?

<@conan_kudo:matrix.org>

16:36:39

I'm skeptical and also formatting :/

<@decathorpe:fedora.im>

16:37:31

I donate my vote to Miro on this one

<@tibbs:fedora.im>

16:37:56

If it reflects reality, sure, but I don't have a firm grip on reality in this case.

<@tibbs:fedora.im>

16:38:57

I'll comment.

<@james:fedora.im>

16:40:36

!topic Open Floor

<@decathorpe:fedora.im>

16:41:02

one of the PRs reminded me of something

<@decathorpe:fedora.im>

16:41:23

apparently there's a "new" way to disable debug packages that nobody told us about?

<@decathorpe:fedora.im>

16:41:33

https://src.fedoraproject.org/rpms/redhat-rpm-config/c/3b9bfae3676327b8a5c1de52f984702220ca8d02?branch=rawhide

<@limb:fedora.im>

16:41:39

Did we need another?

<@decathorpe:fedora.im>

16:42:13

😬

<@conan_kudo:matrix.org>

16:42:47

no, we really didn't :(

<@decathorpe:fedora.im>

16:44:23

yeah I don't know the debug package stuff was changed a fair bit in redhat-rpm-config recently

<@conan_kudo:matrix.org>

16:44:37

the debug package stuff is going to change in rpm too

<@decathorpe:fedora.im>

16:44:41

and I haven't seen *any* public discussions about that

<@conan_kudo:matrix.org>

16:44:48

upstream is finally going to default to debuginfo generation enabled

<@conan_kudo:matrix.org>

16:45:12

and part of that is that we're going to have to clean out our hacky ways to enable it in redhat-rpm-config

<@decathorpe:fedora.im>

16:46:03

once the dust has settled, I guess we should update the packaging guidelines to reflect the new reality?

<@conan_kudo:matrix.org>

16:46:17

!link https://github.com/rpm-software-management/rpm/pull/3040

<@conan_kudo:matrix.org>

16:46:49

!link https://github.com/rpm-software-management/rpm/pull/3036

<@music:fedora.im>

16:47:22

Basically, but I don’t think the phrasing in the PR is very clear. In the README for https://src.fedoraproject.org/rpms/pyproject-rpm-macros, this feature is documented as: Additionally to generated requirements you can supply multiple file names to `%pyproject_buildrequires` macro. Dependencies will be loaded from them: ``` %pyproject_buildrequires requirements/tests.in requirements/docs.in requirements/dev.in ``` In particular, there is nothing special about the filename `requirements.txt`.

<@conan_kudo:matrix.org>

16:47:27

I fully expect that we're going to have to take a huge stick to the debuginfo documentation in our guidelines after this is all said and done

<@tibbs:fedora.im>

16:48:23

One interesting thing I've found recently is that running make in the packaging-committee checkout does not generate documents the same way that the docs site does.

<@james:fedora.im>

16:48:37

This seems bad ... why would the old way not work, why would it not be better to explicitly have it be defined as 0?

<@tibbs:fedora.im>

16:49:03

For example, that weird thing with the review guidelines footnotes: they render just fine in a local build, but not on the docs site. It's not just cosmetic differences.

<@james:fedora.im>

16:49:08

Undefined, to me, implies more "do the default thing" rather than "turn it off"

<@james:fedora.im>

16:49:44

This seems esp. bad if the default is changing upstream.

<@conan_kudo:matrix.org>

16:50:26

our `%global debug_package %{?nil}` thing is a bigger hammer, but at least it does what it's supposed to do

<@tibbs:fedora.im>

16:51:55

Yes, upstream RPM is reworking debug package generation big time and I do expect a lot of the weirder things to break in fun ways.

<@tibbs:fedora.im>

16:52:28

But at this point it's way too early to say.

<@limb:fedora.im>

16:54:47

Best to craft our guidelines on merged PRs.

<@conan_kudo:matrix.org>

16:55:10

I would rather hold off screwing significantly with the documentation until after the upstream work is done.

<@conan_kudo:matrix.org>

16:55:20

Then we'll know whether it'll be part of RPM 4.20 or 4.21

<@conan_kudo:matrix.org>

16:55:22

or RPM 6.0

<@conan_kudo:matrix.org>

16:55:34

(and yes, I really don't know which one it'll be in yet)

<@salimma:fedora.im>

16:56:23

is rpm really jumping from 4.21 to 6.0 next? I guess we do have to skip over 5 huh

<@conan_kudo:matrix.org>

16:56:42

yes

<@conan_kudo:matrix.org>

16:56:50

well I don't know if we're going to have a 4.21

<@conan_kudo:matrix.org>

16:56:54

it might be 4.20 -> 6.0

<@james:fedora.im>

17:02:03

Okay, it's the end of the hour

<@james:fedora.im>

17:02:07

I guess we are done?

<@james:fedora.im>

17:02:50

We can all take a nap, and some meds, while we forget about rpm5 existing.

<@james:fedora.im>

17:03:04

!endmeeting