#meeting-1:fedoraproject.org log

<@yselkowitz:fedora.im>

16:00:47

!startmeeting ELN SIG 10 Mar '26

<@meetbot:fedora.im>

16:00:48

Meeting started at 2026-03-10 16:00:47 UTC

<@meetbot:fedora.im>

16:00:48

The Meeting name is 'ELN SIG 10 Mar '26'

<@yselkowitz:fedora.im>

16:00:51

!meetingname eln

<@meetbot:fedora.im>

16:00:52

The Meeting Name is now eln

<@yselkowitz:fedora.im>

16:00:56

!topic Init process

<@salimma:fedora.im>

16:01:58

!hi

<@zodbot:fedora.im>

16:01:59

Michel Lind (salimma) - he / him / his

<@sgallagh:fedora.im>

16:02:28

!hi

<@zodbot:fedora.im>

16:02:32

Stephen Gallagher (sgallagh) - he / him / his

<@yselkowitz:fedora.im>

16:05:56

pinged some people in channel

<@davide:cavalca.name>

16:06:15

!hi

<@zodbot:fedora.im>

16:06:17

Davide Cavalca (dcavalca) - he / him / his

<@yselkowitz:fedora.im>

16:06:30

ok let's get started

<@yselkowitz:fedora.im>

16:06:38

!topic ELN Hyperscale

<@conan_kudo:matrix.org>

16:07:22

!hi

<@yselkowitz:fedora.im>

16:07:23

iiuc ARK declined to enable btrfs.ko in -modules-internal, so this would require a separate kernel build?

<@zodbot:fedora.im>

16:07:24

Neal Gompa (ngompa) - he / him / his

<@jligon:matrix.org>

16:08:14

!hi

<@zodbot:fedora.im>

16:08:16

No Fedora Accounts users have the @jligon:matrix.org Matrix Account defined

<@conan_kudo:matrix.org>

16:08:21

Yes, but this isn't allowed, so Hyperscale ELN would be fundamentally impossible.

<@yselkowitz:fedora.im>

16:08:33

not allowed meaning what??

<@conan_kudo:matrix.org>

16:09:18

We can't have alternative kernels in fedora, and ark has refused to consider a Hyperscale config overlay

<@salimma:fedora.im>

16:09:18

having two kernel packages I guess?

<@conan_kudo:matrix.org>

16:09:32

We are at an impasse

<@yselkowitz:fedora.im>

16:10:09

on what basis did they refuse to consider a HS config?

<@yselkowitz:fedora.im>

16:10:21

if it doesn't interfere with RHEL or Fedora, why not?

<@conan_kudo:matrix.org>

16:10:52

They don't want to grant a non RH person commit access or code owner rights

<@conan_kudo:matrix.org>

16:11:48

This came up the first time around and there have been no indication they are willing to change on this

<@sgallagh:fedora.im>

16:12:00

So if there was a RH "sponsor", it could happen?

<@yselkowitz:fedora.im>

16:12:12

what more do you need though than a config overlay which enables btrfs.ko and some %centos_hs conditionals in the spec to use that config? once the initial setup is done through an MR, how much more do you need?

<@conan_kudo:matrix.org>

16:12:16

Maybe?

<@conan_kudo:matrix.org>

16:13:29

The overlay has a little bit more, but not much more, but that should be mostly it

<@yselkowitz:fedora.im>

16:15:30

perhaps we should also consider including kmods config together with this, but if the necessary delta can be proposed as an MR in a way that doesn't interfere with RHEL and Fedora configs, then I don't really see reasonable grounds for denial, the question then is how much maintenance this requires on an ongoing basis?

<@yselkowitz:fedora.im>

16:16:00

or iow why do you *need* commit access / code ownership on an ongoing basis?

<@conan_kudo:matrix.org>

16:17:02

Because without that I can't approve MRs for it

<@salimma:fedora.im>

16:17:17

oh folks, I have a conflict at half past, and I do want to bring up a topic - maybe after this

<@conan_kudo:matrix.org>

16:17:25

Or merge them, for that matter

<@yselkowitz:fedora.im>

16:18:29

do we want to pause on this for 15 minutes to give Michel a chance here?

<@conan_kudo:matrix.org>

16:18:50

Sure

<@salimma:fedora.im>

16:19:03

thanks... this involves Neal too (of course it does)

<@yselkowitz:fedora.im>

16:19:03

Michel Lind ☘ UTC floor is yours

<@salimma:fedora.im>

16:19:25

so there's been ongoing conversation in Fedora about improving how CVE bugs are filed and handled

<@salimma:fedora.im>

16:20:27

<@salimma:fedora.im>

16:20:27

any idea, the RH-ers here, if RHEL and Prodsec will find something like PURL useful? that seems close enough to SBOM

<@salimma:fedora.im>

16:20:27

<@salimma:fedora.im>

16:20:27

one idea that came about -- RH ProdSec asked if we can provide some sort of SBOM, since I guess they find it easier to deal with that sort of things

<@salimma:fedora.im>

16:20:27

not going to go into too much details, some public discusisons are here - https://pagure.io/fesco/issue/3512

<@salimma:fedora.im>

16:20:31

one sec will find a link

<@salimma:fedora.im>

16:20:41

https://fossa.com/blog/understanding-purl-specification-package-url/

<@salimma:fedora.im>

16:21:07

because that we can probably auto-generate in multiple package ecosystems from existing metadata, then prodsec can just deal with one spec

<@salimma:fedora.im>

16:21:27

e.g. should be fine for crate(...), python3dist(...), pkgconfig(...), cmake(...)

<@salimma:fedora.im>

16:21:36

I think I got the cmake syntax wrong but you get the idea

<@salimma:fedora.im>

16:21:51

(this was Fabio Valentini's idea)

<@decathorpe:fedora.im>

16:21:52

> As you might expect, both the CycloneDX and SPDX SBOM formats support PURL

<@decathorpe:fedora.im>

16:21:52

<@decathorpe:fedora.im>

16:21:52

notably:

<@yselkowitz:fedora.im>

16:22:29

I'm not sure what this has to do with ELN??

<@decathorpe:fedora.im>

16:22:57

(me either, but I was summoned 👻)

<@salimma:fedora.im>

16:23:57

so this came up in the packaging committee, since it would involve guideline change

<@salimma:fedora.im>

16:24:18

the FPC result is "we don't oppose this but we wonder what other stakeholders would say" and we figure we'd bring this up in ELN and EPEL

<@salimma:fedora.im>

16:24:43

just to see if anyone finds a glaringly obvious issue with using this that we don't see

<@conan_kudo:matrix.org>

16:24:49

As long as we don't some pypi or whatnot, yes

<@conan_kudo:matrix.org>

16:25:18

As long as we don't assume pypi or whatnot, yes

<@yselkowitz:fedora.im>

16:25:35

not having seen the ticket or the article to which you linked, it's not clear to me what the *proposal* even is?

<@conan_kudo:matrix.org>

16:25:42

The only thing I can think of is an inaccurate assumption of source

<@yselkowitz:fedora.im>

16:26:14

maybe you could spell it out in a ticket and we could discuss async before bringing up at a future meeting?

<@salimma:fedora.im>

16:26:16

the proposal is that we autogenerate PURL metadata in RPMs so prodsec can parse that instead of doing their current substring matching on package names

<@salimma:fedora.im>

16:26:37

yeah, sorry, meant to do that but things have been super hectic

<@salimma:fedora.im>

16:27:03

this should probably be a document we can point people to, so we don't copy pasta this all over the place

<@yselkowitz:fedora.im>

16:27:16

yes please, with as much detail as possible

<@salimma:fedora.im>

16:28:12

Conan KudoFabio Valentinilet's follow up on this later, the problem with something that comes up partially in private conversations is I'm not quite sure how much is public on this

<@salimma:fedora.im>

16:28:19

ok, back to the previous topic I guess

<@salimma:fedora.im>

16:28:22

thanks all

<@yselkowitz:fedora.im>

16:28:22

!topic PURL metadata in RPMs

<@yselkowitz:fedora.im>

16:28:40

!action Michel Lind ☘ UTC to file a ticket with details for future discussion

<@yselkowitz:fedora.im>

16:29:03

!topic ELN Hyperscale, resumed

<@yselkowitz:fedora.im>

16:30:21

Conan Kudo 1) is it possible to create a non-conflicting config that could be posted as an MR to kernel-ark, 2) you need a "sponsor" on the RH kernel team who you can work with on HS configs?

<@conan_kudo:matrix.org>

16:31:36

In theory, yes. I already manage our config this way.

<@conan_kudo:matrix.org>

16:32:34

I'd just need to wire up a new overlay folder for Hyperscale and put the configs there, and make it conditionally evaluated (which might be the hard part)

<@conan_kudo:matrix.org>

16:33:31

I'm doubtful of achieving 2) though

<@yselkowitz:fedora.im>

16:33:59

how often would the HS configs need to be touched?

<@conan_kudo:matrix.org>

16:35:12

It doesn't happen very often, maybe once a year or so?

<@yselkowitz:fedora.im>

16:36:10

then this doesn't sound like we're asking for much, a "sponsor" may be an overkill

<@yselkowitz:fedora.im>

16:36:34

you just need someone to accept MRs on a very rare basis (after the initial change)

<@conan_kudo:matrix.org>

16:37:19

Well it also depends on how upstream Linux changes in our bits

<@conan_kudo:matrix.org>

16:37:49

The past is rarely prologue with the kernel

<@yselkowitz:fedora.im>

16:38:04

you mean config options changing and the like?

<@conan_kudo:matrix.org>

16:38:19

Yes

<@conan_kudo:matrix.org>

16:39:06

And I would expect CI tests for the variant build too, so we can catch these

<@yselkowitz:fedora.im>

16:40:25

that may be more of an ask, idk 🤷

<@yselkowitz:fedora.im>

16:40:44

but one step at a time, I think

<@yselkowitz:fedora.im>

16:44:24

if you (yourself or together) can work on the initial MR, perhaps the Meta folks have the leverage to get ARK to accept it?

<@yselkowitz:fedora.im>

16:44:54

if it doesn't interfere with Fedora or RHEL configs, I really don't see reasonable grounds for refusal here

<@yselkowitz:fedora.im>

16:45:39

and if there are issues -- or perhaps even in advance to pave the way -- maybe bring this up at FRCL?

<@conan_kudo:matrix.org>

16:45:42

🤷‍♂️

<@conan_kudo:matrix.org>

16:45:46

I can try

<@yselkowitz:fedora.im>

16:47:42

this may take a bit of work to get in, but Hyperscale is important to Meta so the case should be able to be made for it

<@yselkowitz:fedora.im>

16:48:54

and of course regardless the %centos_hs changes should continue to be upstreamed to rawhide

<@yselkowitz:fedora.im>

16:49:33

for the other packages

<@yselkowitz:fedora.im>

16:49:43

anything else here?

<@davide:cavalca.name>

16:50:31

I'm happy to ask again, but I don't think we have any particular leverage here

<@yselkowitz:fedora.im>

16:51:15

well, one step at a time

<@yselkowitz:fedora.im>

16:51:18

ok 10 minutes left, moving along

<@yselkowitz:fedora.im>

16:51:23

!topic Bootc images

<@yselkowitz:fedora.im>

16:51:32

!link https://github.com/fedora-eln/eln/issues/214

<@yselkowitz:fedora.im>

16:51:58

jligon I left some more feedback on your MR, last I checked there wasn't any change since, any updates?

<@jligon:matrix.org>

16:52:41

I had a busy week. made the changes at the top of the meeting and they failed CI

<@jligon:matrix.org>

16:53:00

really wish I could push without triggering ci there, but unsure how

<@yselkowitz:fedora.im>

16:55:09

obviously I haven't seen those yet, so will have to look later

<@yselkowitz:fedora.im>

16:55:41

but with the changes I proposed, I was able to get it to build and run locally

<@jligon:matrix.org>

16:55:48

its the changes you suggested, and I can ping someone from the bootc side to see what's going on

<@jligon:matrix.org>

16:56:17

I'm clearly not rushing and have plenty to learn

<@yselkowitz:fedora.im>

16:57:37

ok we'll have to follow up in the MR

<@yselkowitz:fedora.im>

16:58:09

!topic Next meeting

<@yselkowitz:fedora.im>

16:58:14

any conflicts next week?

<@sgallagh:fedora.im>

16:59:05

It's St. Patrick's Day. That doesn't really mean anything, other than that the average sobriety level of attendees might be lower than usual.

<@yselkowitz:fedora.im>

16:59:38

that means Michel Lind ☘ UTC may be off

<@yselkowitz:fedora.im>

16:59:56

but the rest of us shouldn't be affected, so we'll keep schedule

<@conan_kudo:matrix.org>

17:00:09

ded

<@yselkowitz:fedora.im>

17:00:21

<@salimma:fedora.im>

17:00:26

it does mean something :)

<@salimma:fedora.im>

17:00:45

checks calendar - yup I'm off

<@jligon:matrix.org>

17:01:08

sure, I just blocked off my time for it

<@jligon:matrix.org>

17:01:26

fewer snakes on an island?

<@yselkowitz:fedora.im>

17:01:28

!info Next meeting is next Tuesday 17 March 12:00 EDT

<@yselkowitz:fedora.im>

17:01:33

!topic Open floor

<@salimma:fedora.im>

17:01:40

there are no snakes on this island :)

<@yselkowitz:fedora.im>

17:01:40

we're already at time, but anything else?

<@bookwar:fedora.im>

17:01:47

While everyone is around - please check https://github.com/fedora-eln/eln/issues/328 We have the prototype for networking (wired and wireless) and package selection screens working in the Installer and we want to demo it soon. But if you have additional requirements, or some specific use cases which we should cover with installer WebUi in ELN - please add them to the ticket

<@yselkowitz:fedora.im>

17:02:07

!link https://github.com/fedora-eln/eln/issues/328

<@yselkowitz:fedora.im>

17:02:43

!info Anaconda team is requesting feedback on preparing WebUI for ELN installer

<@yselkowitz:fedora.im>

17:02:55

thanks bookwar for bringing that up

<@yselkowitz:fedora.im>

17:03:29

anything else?

<@yselkowitz:fedora.im>

17:04:47

in that case, thank you all for contributing, we'll see you in channel

<@yselkowitz:fedora.im>

17:04:49

!endmeeting