#meeting-1:fedoraproject.org log

<@james:fedora.im>

17:00:01

!startmeeting fpc

<@meetbot:fedora.im>

17:00:03

Meeting started at 2026-02-05 17:00:01 UTC

<@meetbot:fedora.im>

17:00:04

The Meeting name is 'fpc'

<@james:fedora.im>

17:00:06

!topic Roll Call

<@gordonmessmer:fedora.im>

17:00:29

!hi

<@zodbot:fedora.im>

17:00:30

Gordon Messmer (gordonmessmer)

<@tibbs:fedora.im>

17:00:32

Howdy.

<@james:fedora.im>

17:00:44

Hey

<@conan_kudo:matrix.org>

17:00:51

!hi

<@zodbot:fedora.im>

17:00:57

Neal Gompa (ngompa) - he / him / his

<@decathorpe:fedora.im>

17:02:36

!hi

<@zodbot:fedora.im>

17:02:37

Fabio Valentini (decathorpe) - he / him / his

<@james:fedora.im>

17:04:43

And then there were only four.

<@james:fedora.im>

17:04:58

!topic FPC PR#1521 https://pagure.io/packaging-committee/pull-request/1521

<@james:fedora.im>

17:05:07

I think this is an easy merge, but wanted to check.

<@carlwgeorge:fedora.im>

17:06:00

!hi

<@zodbot:fedora.im>

17:06:01

Carl George (carlwgeorge) - he / him / his

<@decathorpe:fedora.im>

17:06:02

in general I'm not a fan of neither 1) specifying multiple BRs on one line with `,` nor 2) this strange setup that you need to depend on all these things just to get a working /usr/bin/node ...

<@decathorpe:fedora.im>

17:06:32

point 1) could easily be fixed, 2) probably not

<@yselkowitz:fedora.im>

17:07:09

iiuc it's not all of them, you pick and choose what you need

<@tibbs:fedora.im>

17:07:29

Are they trying really hard to avoid a metapackage?

<@decathorpe:fedora.im>

17:07:52

¯\_(ツ)_/¯

<@carlwgeorge:fedora.im>

17:08:06

weird, this passed fesco with just +2

<@decathorpe:fedora.im>

17:08:30

I'm pretty sure some of the weirdness was added post-approval

<@decathorpe:fedora.im>

17:08:33

the original CP was simpler

<@tibbs:fedora.im>

17:09:46

I don't have a problem with this, really, though if you're supposed to list on the ones that you need, perhaps that should be indicated.

<@carlwgeorge:fedora.im>

17:12:04

i wish more apps took the approach python did, with always versioned source rpms, and the default for a release having unversioned rpms

<@decathorpe:fedora.im>

17:13:16

though the NodeJS setup is supposed to allow you to swap the default to a different version

<@decathorpe:fedora.im>

17:13:16

<@decathorpe:fedora.im>

17:13:16

yeah, from what I can tell, that's basically the "gold standard" for how to handle this

<@james:fedora.im>

17:13:20

I remember a time when people cared about compatibility and we didn't have 666 versions of every module/library.

<@james:fedora.im>

17:13:26

👴

<@decathorpe:fedora.im>

17:13:29

though the NodeJS setup is supposed to allow you to swap the default to a different version (which the Python setup doesn't support)

<@decathorpe:fedora.im>

17:13:29

<@decathorpe:fedora.im>

17:13:29

yeah, from what I can tell, that's basically the "gold standard" for how to handle this

<@james:fedora.im>

17:14:57

Anyway ... do you want to merge and then tweak it ... or refuse to merge before tweaks?

<@carlwgeorge:fedora.im>

17:15:32

can we at least ask for semantic line breaks before merging?

<@decathorpe:fedora.im>

17:15:50

I'm only complaining that this is the way it was implemented ;)

<@decathorpe:fedora.im>

17:15:50

yeah I'm not against documenting what needs to be done.

<@james:fedora.im>

17:15:51

Sure

<@carlwgeorge:fedora.im>

17:16:02

same

<@carlwgeorge:fedora.im>

17:17:00

this `*-bin` naming is weird. arch's aur uses that to indicate packages with pre-compiled binaries.

<@decathorpe:fedora.im>

17:17:43

"-default" or "-unversioned" might have been better ... but at this point, probably too late to fix it for F44

<@conan_kudo:matrix.org>

17:18:06

we can probably fix it after the fact

<@decathorpe:fedora.im>

17:18:09

similar to "python-unversioned-command"

<@james:fedora.im>

17:19:25

!topic FPC PR#1520 https://pagure.io/packaging-committee/pull-request/1520

<@conan_kudo:matrix.org>

17:19:38

but I like the idea of `-default`

<@james:fedora.im>

17:19:44

Gordon Messmer: Want to say anything?

<@gordonmessmer:fedora.im>

17:20:12

https://pagure.io/packaging-committee/issue/1519#comment-1003054

<@gordonmessmer:fedora.im>

17:20:33

only what I added in the comment yesterday. :)

<@decathorpe:fedora.im>

17:20:52

sorry, didn't have a chance to catch up on this issue / PR since FOSDEM

<@james:fedora.im>

17:22:09

Conan Kudo: I assume you are +1 on it?

<@conan_kudo:matrix.org>

17:22:16

yup +1

<@conan_kudo:matrix.org>

17:22:25

we can iterate afterward as needed

<@conan_kudo:matrix.org>

17:22:34

and I think we should probably move his samples to somewhere fedora-y too

<@james:fedora.im>

17:22:39

I'm +1 too, although it feels like the wrong place.

<@carlwgeorge:fedora.im>

17:23:20

perhaps an appendix style page?

<@james:fedora.im>

17:23:34

I think the big problem with that is that if nobody owns it then it'll get lost over time.

<@carlwgeorge:fedora.im>

17:23:37

something like https://docs.fedoraproject.org/en-US/packaging-guidelines/Python_Appendix/

<@conan_kudo:matrix.org>

17:24:46

yes, a shared library appendix probably makes sense logically

<@conan_kudo:matrix.org>

17:25:00

it's not strictly C/C++ specific, even if the current examples are C/C++

<@carlwgeorge:fedora.im>

17:26:01

oh i was thinking specifically about a c/c++ appendix, i'm not sure where a shared appendix would fit structurally

<@conan_kudo:matrix.org>

17:26:14

we have a shared libraries section already

<@conan_kudo:matrix.org>

17:26:22

so just added an appendix page for that is sufficient

<@conan_kudo:matrix.org>

17:26:42

we should probably pull shared libraries into their own page rather than being in the monster page

<@gordonmessmer:fedora.im>

17:26:44

will https://pagure.io/packaging-committee -> a forgejo org? The org can host multiple repos, so that might be a good spot for the "libFoo" examples.

<@conan_kudo:matrix.org>

17:27:10

yes, we haven't figured out what it'll be called yet, but yes

<@decathorpe:fedora.im>

17:27:11

yup

<@conan_kudo:matrix.org>

17:27:48

the C/C++ documentation we have strictly speaking is the wrong kind of docs

<@conan_kudo:matrix.org>

17:27:57

so there's some reorganization we need to do here 😅

<@carlwgeorge:fedora.im>

17:28:17

won't this primarily be c/c++?

<@conan_kudo:matrix.org>

17:28:23

well, no

<@conan_kudo:matrix.org>

17:28:42

you can implement version scripts for capturing symbols for any language that provides an ELF symbol table

<@carlwgeorge:fedora.im>

17:29:06

rephrasing, won't the vast majority of real world implementations be c/c++ programs?

<@conan_kudo:matrix.org>

17:29:26

right now, yes, but in the future I think we'll see more plurality

<@conan_kudo:matrix.org>

17:29:36

it's fine to merge as-is

<@conan_kudo:matrix.org>

17:29:44

but a future step is to clean up and unify the shared library docs

<@conan_kudo:matrix.org>

17:29:52

they are in like 4 different places in the packaging guidelines

<@carlwgeorge:fedora.im>

17:29:53

i think it makes sense to start in the c/c++ doc, and then adjust in the future if that comes to pass

<@conan_kudo:matrix.org>

17:30:23

there's stuff in the monster page, in the c/c++ page, stuff in the versioning page, and so on

<@conan_kudo:matrix.org>

17:30:33

but really, shared libraries are a global concept in fedora

<@carlwgeorge:fedora.im>

17:30:42

i like this and agree it can be merged in the current state, but i would like to see the examples move to a c/c++ appendix page

<@carlwgeorge:fedora.im>

17:31:01

*in the future

<@conan_kudo:matrix.org>

17:31:23

there's stuff in the monster page, in the c/c++ page, stuff in the Python page, and so on

<@gordonmessmer:fedora.im>

17:31:42

sure, I expect reorg and examples for other languages too. but getting the statement up that Fedora wants versioned symbols to improve dep resolution will help me approach more projects with PRs.

<@conan_kudo:matrix.org>

17:31:42

there's stuff in the monster page, in the c/c++ page, stuff in the Python and Rust pages, and so on

<@gordonmessmer:fedora.im>

17:31:48

also, I expect a FAQ

<@gordonmessmer:fedora.im>

17:31:56

The expat maintainer had questions.

<@gordonmessmer:fedora.im>

17:32:19

I chatted with him on Meet for a good while

<@conan_kudo:matrix.org>

17:32:37

the most common languages I expect after C/C++ are Go and Rust

<@conan_kudo:matrix.org>

17:33:12

but again, I don't want to block this on that

<@conan_kudo:matrix.org>

17:33:27

I'm saying we need to make our docs more coherent on this over time

<@james:fedora.im>

17:33:34

I'm going to merge it then. Everyone is encouraged to open more PRs to make things better ;).

<@carlwgeorge:fedora.im>

17:33:34

oh we have a build system docs section, the cmake and meson examples can go in the corresponding pages. i don't see one for automake though.

<@james:fedora.im>

17:34:54

I remember looking at scons a _long_ time ago, and thinking/hoping it'd become a new thing everything used.

<@gordonmessmer:fedora.im>

17:35:10

automake is mentioned, and a practical example is here: https://github.com/gordonmessmer/libFoo/tree/automake

<@carlwgeorge:fedora.im>

17:36:17

sure, i meant in the build systems section of the guidelines, to put the examples in the relevant pages under that

<@carlwgeorge:fedora.im>

17:36:28

e.g. https://docs.fedoraproject.org/en-US/packaging-guidelines/CMake/

<@conan_kudo:matrix.org>

17:36:45

we probably should also start documenting `%conf` and how to use it

<@conan_kudo:matrix.org>

17:36:58

Panu gets cranky when we don't use things we asked for

<@tibbs:fedora.im>

17:38:03

I've used %conf, but the issue as always was EPEL compatibility so people avoid it.

<@conan_kudo:matrix.org>

17:38:21

(`%conf` came out of the discussion about banning `%configure` in `%prep`)

<@tibbs:fedora.im>

17:38:41

I have no idea what actually supports it now.

<@conan_kudo:matrix.org>

17:38:44

(`%conf` came out of the discussion about banning `autoreconf` and `%configure` in `%prep`)

<@conan_kudo:matrix.org>

17:38:48

RHEL 10+

<@conan_kudo:matrix.org>

17:41:11

it was added in RPM 4.18

<@conan_kudo:matrix.org>

17:41:19

alongside `%bcond`

<@tibbs:fedora.im>

17:41:41

So a long time ago I tried to write a section on autoconf but I gut seriously bogged down in basically all the same stuff that people would still argue about.

<@conan_kudo:matrix.org>

17:41:48

alongside `%bcond` improvements

<@tibbs:fedora.im>

17:42:18

But even years ago we recognized that we needed to document autoconf in the same way that we document cmake.

<@conan_kudo:matrix.org>

17:43:30

I wonder if we still have the crazy gnuconfigure macro that regenerates all the autofoo

<@conan_kudo:matrix.org>

17:44:01

mmm seems not

<@james:fedora.im>

17:46:34

!topic Open Floor

<@james:fedora.im>

17:46:46

Been kind of an open floor for a bit, thought I'd make it official ;)

<@james:fedora.im>

17:46:56

Is there anything anyone wants to bring up?

<@mschorm:fedora.im>

17:47:05

!hi

<@zodbot:fedora.im>

17:47:06

None (mschorm)

<@mschorm:fedora.im>

17:47:08

https://pagure.io/packaging-committee/issue/1523

<@conan_kudo:matrix.org>

17:49:35

I think this is straightforward: these are mutually exclusive runtime alternatives

<@conan_kudo:matrix.org>

17:49:47

and it uses the pattern that I should really get around to documenting 😅

<@james:fedora.im>

17:49:51

This seems fine to me ... although this is yet another "we want to provide N versions at once" problem. So would be nice if we could document the "-unversioned" and "-any" and have everyone do the same thing.

<@james:fedora.im>

17:50:03

Those docs. shouldn't hold up this exception though, IMO.

<@conan_kudo:matrix.org>

17:50:13

yeah I think we can ok this

<@conan_kudo:matrix.org>

17:50:32

this is another stick for me to deal with writing this, I think I already said I would a few years ago, but then blech

<@decathorpe:fedora.im>

17:50:42

I'm curious. why "-any"?

<@conan_kudo:matrix.org>

17:50:55

https://pagure.io/packaging-committee/issue/1440

<@decathorpe:fedora.im>

17:50:56

if add virtual provides, why not just make it "mariadb" suffixless?

<@decathorpe:fedora.im>

17:51:51

ah, I see, that's used to denote the default one

<@james:fedora.im>

17:51:55

It basically is ... but there's different parts: server, client, and backup etc.

<@mschorm:fedora.im>

17:52:15

because we already use that.

<@mschorm:fedora.im>

17:52:15

'mariadb10.11' package produces client sub-package called 'mariadb'. The versioned RPMs provides names like 'mariadb10.11' or 'mariadb11.8', and the distribution default version provides the unversioned 'mariadb' name

<@decathorpe:fedora.im>

17:52:21

so this is nodejs from before turned inside out

<@carlwgeorge:fedora.im>

17:53:08

would mariadb10.11-server upgrade to mariadb11.8-server during some fedora system-upgrade?

<@mschorm:fedora.im>

17:54:01

Other teams dealt with same issues, but most did so uncoordinated. And thus we have these various implementations ... like nodejs

<@mschorm:fedora.im>

17:54:01

We invented this to overcame modularity death for MariaDB, MySQL and PostgreSQL - they use pretty much the same code.

<@james:fedora.im>

17:54:49

It's okay, you can blame Conan Kudo ;)

<@mschorm:fedora.im>

17:55:05

AFAIK it should in the case the first is distro default version in fedora X, and the latter is new distro default version in fedora x+1

<@conan_kudo:matrix.org>

17:55:31

and the virtual name can't ever be a real package

<@conan_kudo:matrix.org>

17:55:37

or dnf twists itself into a pretzel

<@carlwgeorge:fedora.im>

17:55:50

it won't upgrade solely because of the virtual provides, you would need obsoletes

<@mschorm:fedora.im>

17:55:55

See https://fedoraproject.org/wiki/Changes/switch_distribution_default_MySQL_from_8.0_to_8.4#How_To_Test for how upgrades work when we switched distro default version in MySQL

<@carlwgeorge:fedora.im>

17:57:33

oh i see, the server subpackage isn't versioned, so it would be an implicit upgrade of the mariadb-server package

<@conan_kudo:matrix.org>

17:57:54

right

<@conan_kudo:matrix.org>

17:58:01

it's the python model instead of the node model

<@james:fedora.im>

17:58:15

!info approved the exception.

<@james:fedora.im>

17:59:07

Also worth noting is that I'll be on PTO on the 19th and 26th ... so if someone else wants to run the meeting, or we can wait until March.

<@conan_kudo:matrix.org>

18:00:49

waiting until March is fine

<@james:fedora.im>

18:01:38

Okay, see ya'll in a month.

<@james:fedora.im>

18:01:41

!endmeeting