#meeting-1:fedoraproject.org log

<@carlwgeorge:fedora.im>

18:01:16

!startmeeting EPEL Steering Committee

<@meetbot:fedora.im>

18:01:18

Meeting started at 2025-04-16 18:01:16 UTC

<@meetbot:fedora.im>

18:01:18

The Meeting name is 'EPEL Steering Committee'

<@carlwgeorge:fedora.im>

18:01:27

!topic howdy

<@salimma:fedora.im>

18:02:00

!hi

<@zodbot:fedora.im>

18:02:01

Michel Lind (salimma) - he / him / his

<@salimma:fedora.im>

18:02:16

oh right no Troy today?

<@dherrera:fedora.im>

18:02:16

!hi

<@elguero:fedora.im>

18:02:17

!hi

<@zodbot:fedora.im>

18:02:21

Diego Herrera (dherrera) - he / him / his

<@zodbot:fedora.im>

18:02:21

None (elguero)

<@davide:cavalca.name>

18:03:01

!hi

<@carlwgeorge:fedora.im>

18:03:02

yeah, he messaged me ask me to run the meeting, he was having some possible hardware issues with his computer

<@zodbot:fedora.im>

18:03:03

Davide Cavalca (dcavalca) - he / him / his

<@nirik:matrix.scrye.com>

18:03:17

morning

<@salimma:fedora.im>

18:03:22

hardware issues affecting even text meetings, ouch

<@rcallicotte:fedora.im>

18:04:25

!hi

<@zodbot:fedora.im>

18:04:26

Robby Callicotte (rcallicotte) - he / him / his

<@carlwgeorge:fedora.im>

18:04:27

!hi

<@zodbot:fedora.im>

18:04:28

Carl George (carlwgeorge) - he / him / his

<@carlwgeorge:fedora.im>

18:05:53

welcome everybody, let's get this show on the road, i think we have quite a few topics to discuss

<@carlwgeorge:fedora.im>

18:06:02

!topic EPEL issues

<@carlwgeorge:fedora.im>

18:06:09

https://pagure.io/epel/issues?tags=meeting&status=Open

<@carlwgeorge:fedora.im>

18:06:40

i can knock out the older one with a quick status update

<@carlwgeorge:fedora.im>

18:06:46

!epel 324

<@zodbot:fedora.im>

18:06:47

**epel #324** (https://pagure.io/epel/issue/324):**EPEL 10 minor version upgrade path**

<@zodbot:fedora.im>

18:06:47

● **Assignee:** carlwgeorge

<@zodbot:fedora.im>

18:06:47

● **Last Updated:** 5 days ago

<@zodbot:fedora.im>

18:06:47

● **Opened:** 4 weeks ago by carlwgeorge

<@zodbot:fedora.im>

18:06:47

<@carlwgeorge:fedora.im>

18:07:29

this is basically done, with some straggler details of mock and the docs

<@carlwgeorge:fedora.im>

18:07:34

!link https://github.com/rpm-software-management/mock/pull/1568

<@carlwgeorge:fedora.im>

18:07:55

!link https://pagure.io/epel/pull-request/326

<@carlwgeorge:fedora.im>

18:08:46

any questions/feedback on those, or should we move on?

<@nirik:matrix.scrye.com>

18:09:00

🎊

<@carlwgeorge:fedora.im>

18:10:16

!epel 328

<@zodbot:fedora.im>

18:10:17

● **Last Updated:** 5 days ago

<@zodbot:fedora.im>

18:10:17

**epel #328** (https://pagure.io/epel/issue/328):**Figure out getting selinux-policy-epel autoinstalled**

<@zodbot:fedora.im>

18:10:17

● **Assignee:** Not Assigned

<@zodbot:fedora.im>

18:10:17

<@zodbot:fedora.im>

18:10:17

● **Opened:** 6 days ago by ngompa

<@carlwgeorge:fedora.im>

18:10:38

Conan Kudo: you around to go over this one?

<@conan_kudo:matrix.org>

18:10:48

yes

<@conan_kudo:matrix.org>

18:10:51

!hi

<@zodbot:fedora.im>

18:10:53

Neal Gompa (ngompa) - he / him / his

<@carlwgeorge:fedora.im>

18:10:59

go ahead

<@conan_kudo:matrix.org>

18:12:08

so basically this situation kind of stinks because we need the policy package installed right after the epel repo is installed because without it, software in epel may fail inscructably

<@conan_kudo:matrix.org>

18:12:32

most methods of automatic installation don't work because dnf's default is to ignore weak dependencies for already installed packages

<@carlwgeorge:fedora.im>

18:13:08

i can confirm that selinux-policy-epel didn't get installed on the upgrade of epel-release for me

<@conan_kudo:matrix.org>

18:13:25

the best I could come up with is supplement the base package on the selinux-policy-epel and hope it triggers when you do an upgrade

<@carlwgeorge:fedora.im>

18:13:49

so tldr, https://src.fedoraproject.org/rpms/epel-release/c/ffadafbe85e0ab364725909a5f8a4558fa5f78ca?branch=epel10 doesn't work

<@carlwgeorge:fedora.im>

18:13:53

right?

<@conan_kudo:matrix.org>

18:14:07

yup

<@carlwgeorge:fedora.im>

18:14:42

instead of hope, can someone prototype this to confirm so we're not just switching to another non-functional approach?

<@nirik:matrix.scrye.com>

18:14:45

couldn't we just make it a requires? or are there actually cases you never want to install selinux policies?

<@carlwgeorge:fedora.im>

18:15:02

if it's a requires then the documented install method doesn't work

<@carlwgeorge:fedora.im>

18:15:21

i.e. `dnf install https://dl.fedoraproject.org/pub/epel/epel-release-latest-10.noarch.rpm`

<@nirik:matrix.scrye.com>

18:15:55

well, no I meant requires in epel-release... instead of reccommends?

<@conan_kudo:matrix.org>

18:16:05

it will always fail then

<@nirik:matrix.scrye.com>

18:16:12

oh right.

<@conan_kudo:matrix.org>

18:16:15

because the repo doesn't exist yet until _after_ epel-release is installed

<@nirik:matrix.scrye.com>

18:16:16

tricky

<@carlwgeorge:fedora.im>

18:16:44

the supplements method seems like a good idea, i'd just like to know if it for sure works the way we want it to

<@salimma:fedora.im>

18:16:50

yeah it will have to be supplement I think

<@salimma:fedora.im>

18:17:08

because then it only gets processed the first time the package is seen, so it will not just fail because the package is not seen

<@salimma:fedora.im>

18:17:23

though how recommends work seems really buggy and we really should ... file an issue to get that fixed anyway?

<@carlwgeorge:fedora.im>

18:18:00

the recommends logic works as advertised, it's a soft dep that is skipped if unavailable

<@conan_kudo:matrix.org>

18:18:19

it used to work the other way, where it always fired regardless, and people complained

<@conan_kudo:matrix.org>

18:18:28

so it was switched

<@nirik:matrix.scrye.com>

18:18:32

you can never win. ;)

<@carlwgeorge:fedora.im>

18:18:37

oh right, the "i uninstalled this, stop re-installing it" thing

<@conan_kudo:matrix.org>

18:18:40

pretty much

<@conan_kudo:matrix.org>

18:18:57

yup

<@carlwgeorge:fedora.im>

18:19:06

so i guess the inverse wasn't done for supplements?

<@conan_kudo:matrix.org>

18:19:17

it was

<@conan_kudo:matrix.org>

18:19:22

but it only applies to installed packages

<@conan_kudo:matrix.org>

18:19:34

technically selinux-policy-epel is not installed yet

<@conan_kudo:matrix.org>

18:20:05

which is why it shouldn't be affected by `exclude_from_weak_autodetect=True` (the current default)

<@carlwgeorge:fedora.im>

18:20:36

can you work up a pr and then test upgrades with the scratch build artifact?

<@carlwgeorge:fedora.im>

18:20:48

or if anyone else wants to volunteer

<@conan_kudo:matrix.org>

18:21:44

I guess I could, is the new stripped selinux-policy package in CentOS Stream yet?

<@conan_kudo:matrix.org>

18:21:55

also, does this affect RHEL 10.0?

<@carlwgeorge:fedora.im>

18:22:58

yes to both

<@conan_kudo:matrix.org>

18:23:25

well at least testing should be straightforward then

<@conan_kudo:matrix.org>

18:23:47

and did selinux-policy-epel make it into EPEL 10.0 too then?

<@carlwgeorge:fedora.im>

18:24:28

confirmed the first, the change was part of selinux-policy-40.13.26-1, current is selinux-policy-40.13.28-1

<@carlwgeorge:fedora.im>

18:25:07

oh good catch, it's missing from 10.0

<@conan_kudo:matrix.org>

18:25:23

I suspected it might (even if I can't check right now)

<@carlwgeorge:fedora.im>

18:25:28

it was first built after the mass branching

<@carlwgeorge:fedora.im>

18:26:18

we can work on testing the upgrade with c10 now, then sort out 10.0 after that

<@conan_kudo:matrix.org>

18:26:48

sounds good

<@salimma:fedora.im>

18:27:18

should we ship a mock config separately just to test that? seems useful

<@carlwgeorge:fedora.im>

18:27:21

see the aforementioned mock pr

<@salimma:fedora.im>

18:27:34

ah, I missed it (multitasking), thanks

<@carlwgeorge:fedora.im>

18:27:42

that adds the epel-z-10 template that would be needed

<@salimma:fedora.im>

18:28:01

nice

<@carlwgeorge:fedora.im>

18:28:18

ok, anything else on this before we move on?

<@carlwgeorge:fedora.im>

18:28:50

i can file the bug to request the 10_0 build of selinux-policy-epel

<@conan_kudo:matrix.org>

18:28:57

sounds good

<@carlwgeorge:fedora.im>

18:29:54

i think that's it for open issues, unless anyone has any they forget to tag for the meeting?

<@jonathanspw:fedora.im>

18:31:01

!hi

<@zodbot:fedora.im>

18:31:02

Jonathan Wright (jonathanspw)

<@jonathanspw:fedora.im>

18:31:08

Here for about 10m

<@carlwgeorge:fedora.im>

18:31:29

!topic old business

<@carlwgeorge:fedora.im>

18:32:03

we've got a pr for something old, let me see if this works...

<@carlwgeorge:fedora.im>

18:32:07

!epel 329

<@zodbot:fedora.im>

18:32:08

Issue querying Pagure: Issue not found

<@carlwgeorge:fedora.im>

18:32:17

guess not

<@carlwgeorge:fedora.im>

18:32:23

!link https://pagure.io/epel/pull-request/329

<@nirik:matrix.scrye.com>

18:33:03

thats not old? but I guess the topic is

<@carlwgeorge:fedora.im>

18:33:18

Michel Lind UTC-6: want to say anything on this one, or just leave it as an fyi for folks to go review

<@carlwgeorge:fedora.im>

18:33:29

yeah old topic, new pr

<@salimma:fedora.im>

18:33:37

I can mention something quick

<@salimma:fedora.im>

18:34:03

so yeah I found two occurences of EPEL Packagers SIG - let me know if I miss any. the page describing what it is, which I now just mark deprecated but not touching further (we can remove it later)

<@salimma:fedora.im>

18:34:38

and also the stalled request process, which I updated to say use the packager SIG that you want, *but* only if they consent to it. With examples pointing to the wikis for Python, Rust and Go SIGs

<@carlwgeorge:fedora.im>

18:35:08

i like it

<@carlwgeorge:fedora.im>

18:35:23

happy to merge after those two small tweaks suggested in the comments

<@salimma:fedora.im>

18:35:30

e.g. for Rust if the package is written in Rust but does not start with rust- the automation to add rust-sig won't work, so you might need to manually request it anyway if the maintainer forgot

<@salimma:fedora.im>

18:35:30

<@salimma:fedora.im>

18:35:30

Python seems to say packagers can add the SIG willy nilly (which seems like a mistake) but.. we should not dictate policy here, the SIGs should

<@salimma:fedora.im>

18:35:50

will look at the tweaks after the meeting, thanks

<@smooge:fedora.im>

18:36:29

I can say that working with the python people in the past... that probably isn't their intent

<@carlwgeorge:fedora.im>

18:36:39

i believe we already voted on the deprecation, so merging this will just close the loop on the paperwork side

<@salimma:fedora.im>

18:37:00

yeah python people tend to expect someone to own the EPEL branch if they're branching

<@carlwgeorge:fedora.im>

18:37:45

https://fedoraproject.org/wiki/SIGs/Python#Python_SIG_FAS_group

<@carlwgeorge:fedora.im>

18:38:40

alright, ready to move on?

<@carlwgeorge:fedora.im>

18:39:11

i have two things for open floor, one small fyi and one can of worms

<@salimma:fedora.im>

18:39:13

yup that link is also in the PR doc

<@salimma:fedora.im>

18:39:14

but yes move on

<@salimma:fedora.im>

18:39:22

can we start with the FYI :)

<@carlwgeorge:fedora.im>

18:39:33

!topic open floor

<@carlwgeorge:fedora.im>

18:39:42

does anyone else want to go first?

<@carlwgeorge:fedora.im>

18:40:43

i'll drop my fyi one, then give others a chance to jump in if they have open floor items

<@carlwgeorge:fedora.im>

18:40:59

!link https://pagure.io/fedora-infra/toddlers/issue/151

<@carlwgeorge:fedora.im>

18:41:25

scm_request_processor now automatically retires the rawhide branch for packages ending with -epel

<@nirik:matrix.scrye.com>

18:41:30

I have a fyi...

<@nirik:matrix.scrye.com>

18:41:37

yeah, that was what mine was gonna be. ;)

<@carlwgeorge:fedora.im>

18:41:45

perfect 😀

<@salimma:fedora.im>

18:42:02

sweet

<@carlwgeorge:fedora.im>

18:42:06

anybody else before my other one?

<@carlwgeorge:fedora.im>

18:43:20

!link https://communityblog.fedoraproject.org/fedora-chooses-forgejo/

<@carlwgeorge:fedora.im>

18:44:32

in case anyone has been living under a rock, the future of fedora git forges is forgejo. they're working on dist-git first, but eventually there will be a general fedora forge too. i saw that pagure.io is now closed for new repos as well, so let's say that's in sunsetting phase now.

<@carlwgeorge:fedora.im>

18:46:36

- random epel issues that cover more than one package

<@carlwgeorge:fedora.im>

18:46:36

not solving this today, but i wanted to ask the question: what do we want to do with pagure.io/epel? right now it's multipurpose.

<@carlwgeorge:fedora.im>

18:46:36

<@carlwgeorge:fedora.im>

18:46:36

- the source for our docs

<@carlwgeorge:fedora.im>

18:46:36

- issues for steering committee topics/votes

<@carlwgeorge:fedora.im>

18:46:36

- issues cle epel team (me and Diego Herrera) issues to track work

<@carlwgeorge:fedora.im>

18:48:00

a long time ago when it looked like fedora was moving to gitlab, i suggested we move to gitlab too. i remember some folks were very against that due to the open core nature, and honestly i'm glad we didn't go that route because i'm not a fan of the gitlab ux.

<@conan_kudo:matrix.org>

18:48:22

my muscle memory for gitlab broke again with their latest UX change

<@carlwgeorge:fedora.im>

18:48:27

not solving this today, but i wanted to ask the question: what do we want to do with pagure.io/epel? right now it's multipurpose.

<@carlwgeorge:fedora.im>

18:48:27

- random epel issues that cover more than one package

<@carlwgeorge:fedora.im>

18:48:27

- issues the cle epel team (me and Diego Herrera) are working on

<@carlwgeorge:fedora.im>

18:48:27

- issues for steering committee topics/votes

<@carlwgeorge:fedora.im>

18:48:27

- the source for our docs

<@carlwgeorge:fedora.im>

18:48:27

<@salimma:fedora.im>

18:48:28

we don't have to decide this today right? or we think it will deteriorate soon

<@conan_kudo:matrix.org>

18:48:28

it annoys me

<@salimma:fedora.im>

18:48:48

if forgejo is ready I suppose we should use forgejo but I don't want to add to their burden right now

<@conan_kudo:matrix.org>

18:48:51

we should wait for the exporter tooling to be written first

<@carlwgeorge:fedora.im>

18:49:24

not today, but i would point out that the ux on pagure.io is already deteriorating, we routinely can't load issues during meetings

<@nirik:matrix.scrye.com>

18:49:39

That was likely not a ux issue.

<@nirik:matrix.scrye.com>

18:49:42

but ai scrapers.

<@conan_kudo:matrix.org>

18:49:46

that's the AI spam load

<@conan_kudo:matrix.org>

18:49:55

it knocks out a lot of stuff

<@conan_kudo:matrix.org>

18:50:01

that's the AI scraper load

<@salimma:fedora.im>

18:50:02

so much ugh for AI

<@nirik:matrix.scrye.com>

18:50:03

(which have been under control for the last week or so... so if you are seeing anything still, please do report it!)

<@carlwgeorge:fedora.im>

18:50:07

not ux in the traditional sense, but if i can't load an issue that is a bad user experience

<@salimma:fedora.im>

18:50:24

but yeah generally pagure.io works better for me than dist-git (where my dashboard never worked because it can't handle hundreds of packages)

<@conan_kudo:matrix.org>

18:50:28

I don't even hate AI/ML personally, but man they really do want me to hate it all

<@carlwgeorge:fedora.im>

18:50:41

lol yeah they're trying

<@salimma:fedora.im>

18:51:05

I do hate it, but I think the scraping is orthogonal and is more "unregulated capitalism is bad"

<@carlwgeorge:fedora.im>

18:51:18

as neal said, waiting for an export tool (and the existense of the fedora general forge instance)

<@salimma:fedora.im>

18:51:18

yay my two pet hates overlap

<@carlwgeorge:fedora.im>

18:51:26

as neal said, waiting for an export tool (and the existence of the fedora general forge instance) is one option

<@nirik:matrix.scrye.com>

18:51:35

Michel Lind UTC-6: thousands you mean? ;)

<@carlwgeorge:fedora.im>

18:52:09

other options are codeberg.org, github.com, and gitlab.com

<@davide:cavalca.name>

18:52:29

I don't really see the point of using an external forge

<@davide:cavalca.name>

18:52:42

It's either forejo or gitlab IMO

<@carlwgeorge:fedora.im>

18:52:58

gitlab is external

<@salimma:fedora.im>

18:52:59

well codeberg runs forgejo - but yeah why bother except to experiment

<@davide:cavalca.name>

18:53:06

Personally leaning towards forejo as epel is a fedora project after all

<@davide:cavalca.name>

18:53:40

For centos stuff we're going to consolidate on gitlab fwiw

<@carlwgeorge:fedora.im>

18:53:48

my thinking with codeberg would be it in theory would be a stepping stone to the fedora general forgejo instance

<@davide:cavalca.name>

18:53:55

As that's where the centos stream development is

<@salimma:fedora.im>

18:54:30

yup

<@davide:cavalca.name>

18:54:30

Do we know when the fedora instance is expected to become available?

<@salimma:fedora.im>

18:54:48

I started using it for the beefy miracle repo just to play around with forgejo too

<@carlwgeorge:fedora.im>

18:54:50

as far as i know, after the dist-git one

<@davide:cavalca.name>

18:55:05

We should probably avoid having multiple forge migrations in a row if we can

<@salimma:fedora.im>

18:55:09

related - will there be anubis or another anti scraping built in from the get go?

<@carlwgeorge:fedora.im>

18:55:10

something about figuring out the harder use case first

<@salimma:fedora.im>

18:55:20

we probably should never launch a replacement forge without one

<@carlwgeorge:fedora.im>

18:55:47

that would be an issue to raise here probably https://codeberg.org/fedora/forgejo-deployment/issues

<@conan_kudo:matrix.org>

18:55:52

we need to get anubis packaged

<@davide:cavalca.name>

18:55:53

I started packaging that fyi. Stuck on some nodejs dependency madness now

<@davide:cavalca.name>

18:56:04

But I can put it somewhere if folks want to help with it

<@conan_kudo:matrix.org>

18:56:12

please do so

<@salimma:fedora.im>

18:56:22

oof. vendoring does not help?

<@salimma:fedora.im>

18:56:32

looks like tgoday's hyperscale is about packaging hells :)

<@conan_kudo:matrix.org>

18:56:33

I can't tell you how often I've sat on my thumbs because I couldn't do anything because the scrapers took everything down

<@carlwgeorge:fedora.im>

18:56:36

we wouldn't be alone on codeberg either, since the forgejo deployment stuff is also being tracked there

<@carlwgeorge:fedora.im>

18:56:46

we could get a repo in https://codeberg.org/fedora

<@conan_kudo:matrix.org>

18:56:47

it made this Fedora release very difficult on me

<@salimma:fedora.im>

18:57:03

yeah. the scraping is getting ridiculously bad

<@carlwgeorge:fedora.im>

18:57:35

on a related note, do we want to keep this multipurpose in the future, or split things up?

<@salimma:fedora.im>

18:57:54

separate docs from the EPSCo repo you mean?

<@carlwgeorge:fedora.im>

18:58:09

yeah i'm thinking about having one repo for docs specifically, and then another one for the steering committee topics/votes

<@salimma:fedora.im>

18:58:10

probably should split it - get a namespace, and two repos

<@conan_kudo:matrix.org>

18:58:14

I think my experience with fesco leans toward not splitting it

<@salimma:fedora.im>

18:58:24

also can we get nested namespaces in forgejo?

<@nirik:matrix.scrye.com>

18:58:33

moving to codeberg and then moving again to a fedora one seems like double the work to me.

<@davide:cavalca.name>

18:58:37

Yeah, I don't think we need to split it

<@davide:cavalca.name>

18:58:43

But also this

<@salimma:fedora.im>

18:58:45

but yeah fesco has this issue where we sometimes forget to look at docs PRs

<@conan_kudo:matrix.org>

18:58:56

we keep forgetting to address issues in the fesco-docs repo because they're not present and visible like the main fesco project

<@salimma:fedora.im>

18:59:02

docs and SIG issues together, any other repo should be split out (e.g. tooling)

<@conan_kudo:matrix.org>

18:59:06

at some point I might actually suggest we move it back into the fesco repo

<@conan_kudo:matrix.org>

18:59:24

probably with the forgejo general forge deployment post DC move

<@salimma:fedora.im>

18:59:30

fwiw fesco is looking at explicitly looking at docs issues before every weekly meeting

<@conan_kudo:matrix.org>

18:59:47

we probably need to file an RFE for CLE engineers to implement it

<@carlwgeorge:fedora.im>

18:59:50

not telling fesco how to work, but at first glance that sounds more like a process issue than an issue with how it's structured

<@conan_kudo:matrix.org>

18:59:54

the Gogs family including forgejo does not support it

<@carlwgeorge:fedora.im>

19:00:18

but the general sig is not the same as the steering committee

<@nirik:matrix.scrye.com>

19:00:18

Michel Lind UTC-6: in case you were curious, I ran the db query that src uses for your home page... you have: 8928 packages that would be listed there. ;)

<@salimma:fedora.im>

19:00:27

we probably don't need it as much if we're on our own instance

<@salimma:fedora.im>

19:00:40

wow, it was probably 3000-ish just last year

<@nirik:matrix.scrye.com>

19:01:06

I think it's a lot of the groups you are in (since it transitively adds those)

<@salimma:fedora.im>

19:01:08

anyway, it's ... time

<@nirik:matrix.scrye.com>

19:01:13

anyhow, yeah, I think nested works...

<@carlwgeorge:fedora.im>

19:01:21

well we're out of time, but i accomplished my goal of putting this out there to get folks thinking about the future

<@carlwgeorge:fedora.im>

19:01:41

as i said before, no decision needed urgently, i'd just like to stop using pagure sooner than later

<@dherrera:fedora.im>

19:03:06

didn't say anything, but my opinion is that we can just wait for the fedora forge to be ready ^^

<@carlwgeorge:fedora.im>

19:03:51

thanks everyone for attending, see y'all next week

<@carlwgeorge:fedora.im>

19:03:56

!endmeeting