<@james:fedora.im>
16:00:38
!startmeeting fpc
<@meetbot:fedora.im>
16:00:40
Meeting started at 2024-04-04 16:00:38 UTC
<@meetbot:fedora.im>
16:00:40
The Meeting name is 'fpc'
<@james:fedora.im>
16:00:43
!topic Roll Call
<@james:fedora.im>
16:01:33
!hi
<@zodbot:fedora.im>
16:01:34
James Antill (james)
<@tibbs:fedora.im>
16:02:21
Hey.
<@tibbs:fedora.im>
16:02:40
Neochat was stuck at "Loading....." for the last 20 minutes and I didn't notice.
<@carlwgeorge:matrix.org>
16:03:10
!hi
<@conan_kudo:matrix.org>
16:03:11
!hi
<@zodbot:fedora.im>
16:03:12
Carl George (carlwgeorge) - he / him / his
<@zodbot:fedora.im>
16:03:13
Neal Gompa (ngompa) - he / him / his
<@limb:fedora.im>
16:07:19
!hi
<@zodbot:fedora.im>
16:07:21
Gwyn Ciesla (limb) - she / her / hers
<@limb:fedora.im>
16:07:33
ISP was ISP-ing...
<@james:fedora.im>
16:09:30
!topic PR#1353 - https://pagure.io/packaging-committee/pull-request/1353
<@james:fedora.im>
16:09:50
I think we can just hit merge on this, any objections?
<@salimma:fedora.im>
16:10:09
!hi
<@zodbot:fedora.im>
16:10:10
Michel Lind (salimma) - he / him / his
<@salimma:fedora.im>
16:11:37
as a Rust packager this looks fine
<@conan_kudo:matrix.org>
16:11:39
yes I think this is good to go James Antill
<@limb:fedora.im>
16:11:56
+1
<@carlwgeorge:matrix.org>
16:11:57
+1
<@salimma:fedora.im>
16:12:15
the documented changes make it much easier to maintain crates
<@james:fedora.im>
16:13:02
Okay, I hit merge ...
<@james:fedora.im>
16:13:20
!topic PR#1351 https://pagure.io/packaging-committee/pull-request/1351
<@conan_kudo:matrix.org>
16:13:43
I'm somewhat uncomfortable about this
<@conan_kudo:matrix.org>
16:14:13
if we're going to do something like this maybe we should just port over the OBS source service mechanism
<@james:fedora.im>
16:14:17
This seems like fallout from the xz mess ... I kind of feel like someone proved the halting problem makes this non-useful a long time ago
<@conan_kudo:matrix.org>
16:14:54
because with source services, they are declarative definitions to run programs/services with defined inputs
<@james:fedora.im>
16:15:13
Conan Kudo: Let's pretend that you know a lot more about OBS than I do ...
<@tibbs:fedora.im>
16:15:22
The basic idea is sound. I swear I typed up a long comment yesterday at home but I must not have ubmitted it.
<@limb:fedora.im>
16:15:44
I'd be more comfortable with SHOULD
<@tibbs:fedora.im>
16:15:54
The problem is in the details. What's going to substitute %{version} and such in this "pesudo-shell script"
<@conan_kudo:matrix.org>
16:15:56
I'd rather go with MAY
<@limb:fedora.im>
16:15:59
Or maybe COULD or MIGHTBENICE
<@carlwgeorge:matrix.org>
16:16:01
this seems fine to me. even if we want something more robust in the future, this is better than the status quo.
<@limb:fedora.im>
16:16:06
OUGHT
<@tibbs:fedora.im>
16:16:16
Does that exist? If not then there's no point in merging it anyway.
<@conan_kudo:matrix.org>
16:16:20
COULDA
<@nhanlon:beeper.com>
16:16:30
MAYHAPS
<@limb:fedora.im>
16:16:39
Oooh
<@carlwgeorge:matrix.org>
16:16:55
one of my packages has something like this, and i settled on the script requiring a single argument of the upstream version number
<@tibbs:fedora.im>
16:17:04
Right now this appears to be just a brainstorming proposal. I'm happy with saying "flesh it out", but there are still questions about the details.
<@tibbs:fedora.im>
16:17:35
Right, why not pass version, release and name in the environment or as positionals?
<@tibbs:fedora.im>
16:18:12
That at least makes it semi-useful without having fedpkg learn how to make those substitutions in the script itself.
<@carlwgeorge:matrix.org>
16:18:49
ideally name would be hard coded as this would be a per repo script and the name wouldn't be changing, and i don't think the relelase field would be necessary
<@tibbs:fedora.im>
16:19:04
Bottom line for me is that I think we definitely should have a standard for doing this that could potentially hook into automation at some point.
<@tibbs:fedora.im>
16:19:33
Packages don't have to follow it, but if someone wants to do this we should have at least suggestions on how it should be done.
<@tibbs:fedora.im>
16:19:56
Even picking a name for the script puts us in a better place than we are now.
<@carlwgeorge:matrix.org>
16:19:59
the first line does have a lowercase "should", perhaps it should just be capitalized
<@salimma:fedora.im>
16:20:05
agreed
<@limb:fedora.im>
16:20:34
There's also a must on line 22.
<@limb:fedora.im>
16:20:41
Fooling consistency, yadda yadda
<@carlwgeorge:matrix.org>
16:20:43
yeah i just noticed that
<@tibbs:fedora.im>
16:20:55
But I just can't get on board with this odd requirement for something to substitute in %name, %version and %release.
<@carlwgeorge:matrix.org>
16:21:02
probably drop that one for clarity, but i did understand it to be in the context of "if you do this, you must..."
<@conan_kudo:matrix.org>
16:21:13
mmm
<@salimma:fedora.im>
16:24:54
yeah. it makes sense that if you use the new-style format you use it in the prescribed way
<@salimma:fedora.im>
16:25:02
not sure what's a good way to word that though
<@salimma:fedora.im>
16:25:22
I guess "Should you do this, you must do it this way" - so make the linkage explicit?
<@james:fedora.im>
16:26:37
Eh, I feel like if we pass something like this we'll need a bunch of volunteers (and/or automation) to look at compliance in packages.
<@james:fedora.im>
16:28:15
I don't think I like that ... it should be more like "If you can't/won't do this ... please let XYZ know why"
<@james:fedora.im>
16:29:35
But, I'd feel much better if we could move to an end goal of the scripts being really small and/or one of a few known scripts ... then change it out to be "here is a config. file for how the source is generated, in one of N known ways"
<@tibbs:fedora.im>
16:29:58
I think a first pass should have minimal requirements. Just if the script carries a name like XXX then it MUST follow the prescribed format so that automation can make assumptions.
<@tibbs:fedora.im>
16:30:18
But you don't have to have a script at all, and if you do it doesn't have to carry a particular name.
<@james:fedora.im>
16:30:42
I do sympathize with Carl George 's point that a script would be better than the current state ... but feel like if we are prepared to do a bunch of work we should look for an end goal that's better.
<@salimma:fedora.im>
16:30:45
right. so once we have automation we can make the assumption for scripts that match this particular naming convention
<@james:fedora.im>
16:31:35
I guess I'm fine with the minimal one, then we can at least see where we are in a year
<@carlwgeorge:matrix.org>
16:31:59
`fedpkg download-new-upstream-source-that-definitely-doesnt-have-an-xz-style-backdoor`
<@james:fedora.im>
16:32:03
And we don't have to do much :)
<@conan_kudo:matrix.org>
16:32:09
oh dear
<@nhanlon:beeper.com>
16:34:57
| clamscan ?
<@nhanlon:beeper.com>
16:35:19
just woodshedding ;)
<@james:fedora.im>
16:38:59
Anyone else want to post a comment on the PR about what we said?
<@tibbs:fedora.im>
16:39:06
I will add my thoughts in the ticket.
<@conan_kudo:matrix.org>
16:39:54
I'll put my thoughts in there too
<@conan_kudo:matrix.org>
16:40:04
assuming it isn't covered by others :)
<@tibbs:fedora.im>
16:40:58
I guess that right now there is more attention on doing something fancier than just downloading supplied tarballs.
<@tibbs:fedora.im>
16:41:44
I'm surprised there isn't more effort going into just working with git repositories directly. I recall a bunch of work being put into that many years ago.
<@james:fedora.im>
16:42:32
Yeh, IIRC there was a push at one point to have exploded source in distrogit
<@tibbs:fedora.im>
16:43:42
My swiss cheese memory is suggesting to me that Colin Walters was involved in that effort. I have no idea why my brain served up that name, but there it is.
<@carlwgeorge:matrix.org>
16:44:02
yes, source-git, and it was not well received
<@conan_kudo:matrix.org>
16:44:11
source-git wasn't about pristine sources either
<@conan_kudo:matrix.org>
16:44:15
which doesn't help any of this
<@conan_kudo:matrix.org>
16:44:33
a lot of the aims around that was oriented around forking things, which is not what most of us want
<@james:fedora.im>
16:45:04
"How many Fedora packagers does it take to change a lightbulb?"
"CHANGE?!?!?"
<@carlwgeorge:matrix.org>
16:45:33
i don't remember colin being involved, but he could have been
<@limb:fedora.im>
16:45:49
LIGHTBULG-NG
<@carlwgeorge:matrix.org>
16:45:50
it was mostly from the team behind packit, and packit has native support for it
<@limb:fedora.im>
16:45:56
LIGHTBULB-NG
<@carlwgeorge:matrix.org>
16:46:02
https://packit.dev/source-git
<@tibbs:fedora.im>
16:46:37
What I'm thinking of predated packit by some time, I think. Unless packit was being developed in secret for a decade.
<@tibbs:fedora.im>
16:46:48
Anyway, it's certainly off-topic.
<@conan_kudo:matrix.org>
16:47:06
rpm-gitoverlay was colin
<@conan_kudo:matrix.org>
16:47:13
he also wanted us to have a monorepo :/
<@conan_kudo:matrix.org>
16:47:22
oh and everything in kubernetes too
<@conan_kudo:matrix.org>
16:47:37
he lives in a different world 🤷♂️
<@tibbs:fedora.im>
16:48:10
I've found that many good ideas are ignored because they are wrapped up with rather bad ideas and presented as a package.
<@tibbs:fedora.im>
16:49:26
We are in a different place then we were 15 years ago and careful automation has proven to work well, such as with autochangelog and such. So maybe it's time to dig through old proposal to find the good ideas. But I'm certainly not the one to do it.
<@tibbs:fedora.im>
16:50:18
Anyway, before we run out of time, does anyone know what's going on with MPI?
<@tibbs:fedora.im>
16:50:54
I'm not familiar with the person who is asking us to change the MPI guidelines, and I'm not sure if those changes have buy-in from the rest of the scitech packagers.
<@james:fedora.im>
16:51:01
!topic Open Floor
<@james:fedora.im>
16:51:11
Jason ティビツ: I assume you mean: https://pagure.io/packaging-committee/issue/1345 ?
<@tibbs:fedora.im>
16:51:32
And https://pagure.io/packaging-committee/pull-request/1335
<@tibbs:fedora.im>
16:52:12
And I guess https://pagure.io/packaging-committee/issue/1320
<@tibbs:fedora.im>
16:53:11
I'm just not sure what to do with it. It deserves consideration but we aren't experts here, and so we could just accept what he wants to push as long as its sane. But there is a community around the scitech stuff that should probably be involved.
<@james:fedora.im>
16:54:12
yeh
<@conan_kudo:matrix.org>
16:54:56
all of my knowledge of this stuff is horrifically out of date
<@conan_kudo:matrix.org>
16:55:03
I haven't done anything with this in over a decade
<@conan_kudo:matrix.org>
16:55:14
so I really don't know how to advise here
<@decathorpe:fedora.im>
16:56:09
hey 👋🏼 sorry that I couldn't attend today, I'm not at home and only connected by phone.
<@james:fedora.im>
16:56:12
yeh, I didn't have much knowledge anyway ... but it was all from over a decade ago.
<@tibbs:fedora.im>
16:57:18
I will just say that somewhere.
<@james:fedora.im>
16:57:23
I guess we could just ping a bunch of the maintainers of MPI packages?
<@conan_kudo:matrix.org>
16:58:27
do we have any MPI maintainers?
<@conan_kudo:matrix.org>
16:58:38
or is it one of those "stack maintained" things?
<@tibbs:fedora.im>
16:59:35
I think it's changed a bit over time and now we have new people doing it. I don't know how much coordination there still is.
<@limb:fedora.im>
17:00:12
I maintain one of the Python bits and they haven't needed much from me.
<@james:fedora.im>
17:03:37
Yeh, it might well be one of those things that doesn't change much so nobody is deeply interested anymore
<@james:fedora.im>
17:04:00
Anyway ... we are three minutes over time ... anyone have anything before I end the meeting?
<@tibbs:fedora.im>
17:05:02
Not from me. I will go ahead and merge the other semantic breaks thing I did for GAP.
<@tibbs:fedora.im>
17:05:37
Maybe next week we can take up the "writing to /var" thing.
<@tibbs:fedora.im>
17:05:47
If we want a fun and exciting topic.
<@james:fedora.im>
17:06:18
I'll be out next week ... so that sounds like a great topic for ya'll ;)
<@james:fedora.im>
17:06:24
!endmeeting