<@james:fedora.im>
16:00:22
!startmeeting fpc
<@meetbot:fedora.im>
16:00:23
Meeting started at 2025-07-17 16:00:22 UTC
<@meetbot:fedora.im>
16:00:23
The Meeting name is 'fpc'
<@james:fedora.im>
16:00:26
!topic Roll Call
<@salimma:fedora.im>
16:00:29
!hi
<@zodbot:fedora.im>
16:00:30
Michel Lind (salimma) - he / him / his
<@limb:fedora.im>
16:00:36
!hi
<@zodbot:fedora.im>
16:00:37
Gwyn Ciesla (limb) - she / her / hers
<@decathorpe:fedora.im>
16:00:48
!hi
<@zodbot:fedora.im>
16:00:49
Fabio Valentini (decathorpe) - he / him / his
<@james:fedora.im>
16:00:58
!hi
<@zodbot:fedora.im>
16:00:59
James Antill (james)
<@carlwgeorge:fedora.im>
16:01:10
!hi
<@zodbot:fedora.im>
16:01:13
Carl George (carlwgeorge) - he / him / his
<@conan_kudo:matrix.org>
16:05:21
!hi
<@zodbot:fedora.im>
16:05:23
Neal Gompa (ngompa) - he / him / his
<@james:fedora.im>
16:06:16
Internet and no power ... is pretty impressive
<@james:fedora.im>
16:06:37
!topic FPC#1468 https://pagure.io/packaging-committee/issue/1468
<@conan_kudo:matrix.org>
16:07:27
I wish :)
<@decathorpe:fedora.im>
16:07:32
fwiw I'm +1 to this suggestion from the ticket
<@decathorpe:fedora.im>
16:07:32
<@decathorpe:fedora.im>
16:07:32
> I suppose we could just remove "TEX" (sic) from the list and then note that TeX fonts are not covered.
<@salimma:fedora.im>
16:08:15
I'm +1 too
<@conan_kudo:matrix.org>
16:08:20
+1
<@limb:fedora.im>
16:08:20
I'm also in favor of making it all make sense....
<@james:fedora.im>
16:08:21
Yeh, I'm fine to just shrug and say it's fine ... but it feels like we should change something.
<@james:fedora.im>
16:08:26
As tibbs said in his comment.
<@salimma:fedora.im>
16:08:33
I think the texlive namespace should trump the font namespace in this case, since it has tooling implications
<@conan_kudo:matrix.org>
16:09:07
at least in opensuse and I think also in mageia, you are supposed to have the font type in the package name
<@carlwgeorge:fedora.im>
16:09:40
+1
<@conan_kudo:matrix.org>
16:09:40
or I might have that backwards
<@conan_kudo:matrix.org>
16:09:42
that might be mageia
<@conan_kudo:matrix.org>
16:10:08
oh blech, the guidelines are there but nobody cares
<@decathorpe:fedora.im>
16:10:36
🙌
<@limb:fedora.im>
16:10:45
whoop whopp
<@limb:fedora.im>
16:11:09
Plus I feel like the fonts ecosystem was being majorly updated and then....wasn't?
<@limb:fedora.im>
16:11:21
(apropos font macro warnings)
<@decathorpe:fedora.im>
16:11:21
it was
<@conan_kudo:matrix.org>
16:11:40
yeah, there was an attempt by nim several years ago which didn't really take hold
<@limb:fedora.im>
16:11:56
That's what I was thinking.
<@decathorpe:fedora.im>
16:12:04
well ... the current guidelines require it, don't they?
<@james:fedora.im>
16:17:34
Do we want to try to change any of the guidelines, or just shrug and say this ticket is fine and TeX has an exception?
<@limb:fedora.im>
16:18:10
I don't feel like we have a solid concept for a rewrite, so probably shrug.
<@limb:fedora.im>
16:18:34
If I'm wrong, someone can write that up. :)
<@salimma:fedora.im>
16:18:50
yeah let's not block this on the rewrite
<@carlwgeorge:fedora.im>
16:19:10
yeah, exception for now, and throw the docs rewrite on the pile of "it would be nice for someone to get around to a pr for this"
<@decathorpe:fedora.im>
16:19:28
which file format are TeX fonts shipped in?
<@decathorpe:fedora.im>
16:20:46
and ... this is *already* in the current font guidelines:
<@decathorpe:fedora.im>
16:20:46
<@decathorpe:fedora.im>
16:20:46
> Support for other font systems, for specific applications, non-OpenType font formats, bulky documentation, TEX, CSS, or JSON files… MUST be split in separate non-font packages, that SHOULD install outside /usr/share/fonts, and MUST NOT use <something>-fonts naming.
<@salimma:fedora.im>
16:21:48
one sec (just realizing my main machine does not currently have tex installed)
<@salimma:fedora.im>
16:23:59
tfm and mf, looks like
<@decathorpe:fedora.im>
16:24:41
so I'd say this is covered by existing guidelines 🤷
<@decathorpe:fedora.im>
16:24:56
unless I'm even more clueless about fonts than I think I am :)
<@james:fedora.im>
16:25:20
Well the last part about "MUST NOT" is wrong.
<@salimma:fedora.im>
16:25:27
<@salimma:fedora.im>
16:25:27
1 txt
<@salimma:fedora.im>
16:25:27
```
<@salimma:fedora.im>
16:25:27
```
<@salimma:fedora.im>
16:25:27
23 tfm
<@salimma:fedora.im>
16:25:27
1 texlive-latex-fonts
<@salimma:fedora.im>
16:25:27
28 mf
<@salimma:fedora.im>
16:25:27
2 latex-fonts
<@salimma:fedora.im>
16:25:27
⬢ [fedora-packaging] ❯ for f in (rpm -ql texlive-latex-fonts | grep latex-fonts ); basename $f | cut -d. -f2; end | sort | uniq -c
<@decathorpe:fedora.im>
16:26:54
doesn't this mean that texlive fonts not using "*-fonts" naming is correct? ... or are they using "texlive-*-fonts" and that's the "problem"?
<@decathorpe:fedora.im>
16:27:09
doesn't this mean that texlive fonts not using "\*-fonts" naming is correct? ... or are they using "texlive-\*-fonts" and that's the "problem"?
<@james:fedora.im>
16:28:44
Yeh, the "problem" is that `texlive-latex-fonts` ends in `-fonts` but doesn't install system fonts into `/usr/share/fonts`
<@decathorpe:fedora.im>
16:29:24
ugh, ok, now I get it ...
<@james:fedora.im>
16:30:01
!topic FPC#1470 https://pagure.io/packaging-committee/issue/1470
<@limb:fedora.im>
16:30:28
And of course some TeX fonts *are* available for general use, and sometimes I do...cuz they're neat....
<@salimma:fedora.im>
16:31:05
ah yes, this one is intersting
<@decathorpe:fedora.im>
16:32:04
I would direct them at https://src.fedoraproject.org/rpms/redhat-rpm-config/blob/rawhide/f/buildflags.md and say "do something similar" ...
<@james:fedora.im>
16:32:35
I'm pretty sure I don't want people to do this as hacky one off things.
<@decathorpe:fedora.im>
16:33:22
yes, I meant, direct the reporter of the ticket there, not individual people :)
<@decathorpe:fedora.im>
16:33:33
so ... something like `%undefine _strip_static_debuginfo`, or `%global _strip_static_debuginfo 1` with 0 as default?
<@james:fedora.im>
16:33:54
I can live with "static libs have debuginfo" or "statis libs don't have debuginfo" ... and even with "we have 2x the packages for debug or not" ... but randomly turning it off and on for each static lib in packages seems terrible.
<@decathorpe:fedora.im>
16:34:18
this already happens though ...
<@james:fedora.im>
16:34:37
Well if people could stop, that'd be great ;)
<@decathorpe:fedora.im>
16:34:57
for example, the package for the Rust standard library needs to turn stripping of static libs *off* right now because otherwise you get no debuginfo for the standard library in *any* Rust packages :)
<@james:fedora.im>
16:35:57
Okay ... but what's the argument for hacking around it after we change the default?
<@decathorpe:fedora.im>
16:37:25
that argument goes away for *this* case
<@decathorpe:fedora.im>
16:37:44
but there will be packages where maintainers will want to keep the *current* behaviour (i.e. strip static libraries)
<@decathorpe:fedora.im>
16:39:23
(in other words: it doesn't matter whether "strip" or "don't strip" will be the default, some packages will need to opt out / opt in regardless)
<@tibbs:fedora.im>
16:39:28
Ugh; I'm running late today.
<@james:fedora.im>
16:41:24
I guess my question is why do they need to strip it? I can understand "you can't debug any rust app. with the current default, so we have to workaround it" ... but "our static lib. looks too big with the new default, so we have to workaround it" doesn't convince me the same way.
<@salimma:fedora.im>
16:46:28
yeah since the static lib is only installed by those who need to compile against it, right?
<@salimma:fedora.im>
16:46:45
I should check if Debian strips their static libs
<@decathorpe:fedora.im>
16:47:24
apparently one of the things that would be affected is llvm-devel, growing from a few hundred MB to a few GB or something
<@decathorpe:fedora.im>
16:47:36
and pulling in llvm-devel is not *that* rare for packages
<@decathorpe:fedora.im>
16:48:26
(not sure if the static archives could just not get pulled in by llvm-devel by default)
<@james:fedora.im>
16:49:48
I mean ... I remember when a few GB was a lot. But most Fedora users don't need to install llvm-devel.
<@decathorpe:fedora.im>
16:50:09
but koji / COPR builds do
<@decathorpe:fedora.im>
16:51:05
I'm not saying changing the default is a bad idea (in fact I think it's a good idea!). but there *will* be packages that will need to opt in to stripping .a files.
<@james:fedora.im>
16:51:09
Is build disk space a problem?
<@decathorpe:fedora.im>
16:51:23
more like ... network transfer and buildroot installation time ;)
<@james:fedora.im>
16:51:57
I mean ... I remember when network transfering a few GB was a lot.
<@decathorpe:fedora.im>
16:52:16
I remember vividly because it's *still* a lot for most of the world
<@james:fedora.im>
16:53:20
I'm pretty sure it's not a problem moving the data around between koji builders and the repos. they install from.
<@decathorpe:fedora.im>
16:53:56
(again, I'm not saying that the llvm thing this is a *good* case, just that there *will* be cases)
<@james:fedora.im>
16:56:46
I think our policy on build flags is that you can get an exception, if you really need it? I would be fine on saying the same for removing debug info. from static libs.
<@conan_kudo:matrix.org>
16:57:17
if it's macroized as a knob, then it should be fine either way
<@james:fedora.im>
16:57:53
Kind of ... But I really don't want people to randomly turn it off because they feel like it.
<@conan_kudo:matrix.org>
16:58:17
I don't think most people will
<@decathorpe:fedora.im>
16:58:17
is "getting an exception" even still required here? I thought "I know what I'm doing" is the current policy 🙃
<@decathorpe:fedora.im>
16:58:17
<@decathorpe:fedora.im>
16:58:17
> get an exception, if you really need it?
<@conan_kudo:matrix.org>
16:58:30
it is 🙃
<@conan_kudo:matrix.org>
16:59:02
the only exception to that until recently was compilers
<@james:fedora.im>
16:59:09
Our current policy on build flags is "here they are, but feel free to not use them if you feel like it?"
<@conan_kudo:matrix.org>
16:59:46
well we sort of enforce using the flags by ensuring the rpm build is broken if you're not really using them
<@conan_kudo:matrix.org>
17:00:04
"ensuring" is a bit strong of a word, it's more like the cascade of policies basically make it work that way
<@decathorpe:fedora.im>
17:00:22
and ... mostly by accident
<@conan_kudo:matrix.org>
17:00:32
yeeaaah
<@james:fedora.im>
17:00:59
Maybe you are thinking about `_pkg_extra_cflags` ... which isn't really the same as changing them how you want. Although I guess someone could add flags to turn off defaults.
<@decathorpe:fedora.im>
17:02:01
does even everybody check that their CMake build system doesn't override weird things in CFLAGS? ...
<@decathorpe:fedora.im>
17:02:16
you mostly only notice when upstream does something *really really* stupid
<@conan_kudo:matrix.org>
17:02:42
I do at least
<@conan_kudo:matrix.org>
17:02:57
but I don't think most people do that with any build script
<@decathorpe:fedora.im>
17:03:00
still, all you can do is file bugs against the packages that do this, and most maintainers just meh out
<@james:fedora.im>
17:03:45
Joy
<@james:fedora.im>
17:04:19
Anyway, we are 4 minutes over time now ... anythign anyone wants to say before I end the meeting?
<@conan_kudo:matrix.org>
17:04:29
nope
<@james:fedora.im>
17:05:04
!endmeeting