#meeting-1:fedoraproject.org log

<@james:fedora.im>

16:00:38

!startmeeting fpc

<@meetbot:fedora.im>

16:00:39

Meeting started at 2024-07-18 16:00:38 UTC

<@meetbot:fedora.im>

16:00:39

The Meeting name is 'fpc'

<@james:fedora.im>

16:00:42

!topic Roll Call

<@salimma:fedora.im>

16:01:03

!hi

<@zodbot:fedora.im>

16:01:05

Michel Lind (salimma) - he / him / his

<@conan_kudo:matrix.org>

16:01:13

!hi

<@zodbot:fedora.im>

16:01:15

Neal Gompa (ngompa) - he / him / his

<@james:fedora.im>

16:01:17

!hi

<@zodbot:fedora.im>

16:01:18

James Antill (james)

<@jsteffan:fedora.im>

16:02:57

!hi

<@zodbot:fedora.im>

16:02:59

Jonathan Steffan (jsteffan)

<@limb:fedora.im>

16:03:16

!hi

<@zodbot:fedora.im>

16:03:18

Gwyn Ciesla (limb) - she / her / hers

<@salimma:fedora.im>

16:09:59

wondering if Matrix is glitching again or we've been silent for 6 mins

<@james:fedora.im>

16:10:44

We are usually quiet at the start, hoping for 5+ people to turn up from the committee

<@james:fedora.im>

16:10:51

!topic Open Floor

<@james:fedora.im>

16:11:57

!link https://pagure.io/packaging-committee/pull-request/1368

<@james:fedora.im>

16:12:18

This got rebased, but I'm not sure we should approve it without 5

<@james:fedora.im>

16:12:57

Would be happy to if the packagers affected would come into the ticket and say it seems fine to them.

<@jsteffan:fedora.im>

16:13:08

yeah, i emailed all of the contributors for openxr, vulkan, and mesa directly but still didn't see any feedback

<@jsteffan:fedora.im>

16:13:24

yeah, i emailed all of the maintainers for openxr, vulkan, and mesa directly but still didn't see any feedback

<@james:fedora.im>

16:14:03

Michel Lind 🎩: Can you give it a read and post a comment in the PR?

<@salimma:fedora.im>

16:14:08

oh, this does not affect my use case though

<@james:fedora.im>

16:14:16

Doesn't need to be during the meeting.

<@jsteffan:fedora.im>

16:14:31

it should cover your use-case.. so what is your use case Michel Lind 🎩 so i can add it

<@jsteffan:fedora.im>

16:14:44

i was hoping to cover all unversioned SONAMEs

<@salimma:fedora.im>

16:14:47

for mine it's ... a tool maintained by a project that they unfathomably chose to fork for another project, and when building the fork that tool is installed and includes an unversioned so :(

<@conan_kudo:matrix.org>

16:14:48

<@conan_kudo:matrix.org>

16:14:48

the only case I'm slightly worried about is:

<@conan_kudo:matrix.org>

16:14:48

> The DSO is not linked against and is loaded at runtime by the implementing application (i.e. `dlopen()`)

<@conan_kudo:matrix.org>

16:15:03

this implies that it's okay for it to exist in `/usr/lib64`

<@jsteffan:fedora.im>

16:15:07

there just specific guidance for vulkan and openxr because that is where i've been focused

<@conan_kudo:matrix.org>

16:15:10

which... I'm very wary about

<@salimma:fedora.im>

16:15:27

so I think in my case (zeek and spicy) libspicy should just either be fixed to version its soname or I should package the standalone spicy and force zeek to use that

<@james:fedora.im>

16:15:45

My understanding is that is kind of true about vulkan/openxr

<@salimma:fedora.im>

16:15:56

> - This means that they are not located directly in /usr/lib or /usr/lib64,

<@salimma:fedora.im>

16:15:56

this seems clear though

<@salimma:fedora.im>

16:15:56

<@james:fedora.im>

16:16:06

In that nobody will ever convince them otherwise, and nobody wants to say no.

<@conan_kudo:matrix.org>

16:16:18

the writing for the conditions makes me think they are not a combination of conditions, and only any one of them have to be true

<@jsteffan:fedora.im>

16:16:33

right before that, a condition for the default policy is:

<@jsteffan:fedora.im>

16:16:33

* The DSO is located in a private directory (i.e. not located directly in /usr/lib[64] or in another directory listed as a library path for the linker)

<@conan_kudo:matrix.org>

16:16:45

the way I would prefer to read it is that all three conditions need to be true

<@michel:one.ems.host>

16:16:49

drats chat.fedoraproject.org went down on me

<@jsteffan:fedora.im>

16:17:17

this was the intent. more than willing to try to make it more clear.

<@salimma:fedora.im>

16:17:24

one of my message is lost but yeah I was just saying this

<@conan_kudo:matrix.org>

16:18:24

yeah, I think this is a wordsmithing issue...

<@conan_kudo:matrix.org>

16:18:24

> The following conditions MUST be true in which DSOs are allowed to remain unversioned.

<@conan_kudo:matrix.org>

16:18:24

<@conan_kudo:matrix.org>

16:18:24

This should be be written as:

<@conan_kudo:matrix.org>

16:18:24

<@conan_kudo:matrix.org>

16:18:24

> There are a few conditions that need to exist in which DSOs are allowed to remain unversioned.

<@conan_kudo:matrix.org>

16:18:24

<@james:fedora.im>

16:18:41

Ahh, I see, upon better reading you want to change how vulkan/etc work?

<@jsteffan:fedora.im>

16:19:03

okay, i give me a sec to propose a quick re-write of the default case to use the RFC language

<@conan_kudo:matrix.org>

16:19:43

I am confused about adding ldconfig configuration snippets

<@conan_kudo:matrix.org>

16:19:54

that implies vulkan loader modules are actually cached by ldconfig

<@conan_kudo:matrix.org>

16:20:29

I do get the special exception part, it's written there, but I think I missed something in all the conversations about this

<@jsteffan:fedora.im>

16:24:21

okay, i've updated the PR with more clear RFC language for the default case. i hope that helps

<@jsteffan:fedora.im>

16:25:30

vulkan and openxr loader (i.e. vulkan-loader and openxr-loader) have been specifically designed to use the default linker to find things and it was an intentional design decision

<@jsteffan:fedora.im>

16:26:06

the vulkan-loader and openxr-loader act as a trampoline in the load path, with implementing applications linking against the loader directly

<@jsteffan:fedora.im>

16:27:36

airlied commented about vulkan drivers on https://pagure.io/packaging-committee/issue/1365 and they do need to visible

<@james:fedora.im>

16:28:54

I think it looks fine to me.

<@jsteffan:fedora.im>

16:30:27

the other thing that i've kept in mind is https://gitlab.collabora.com/vivek/libcapsule/-/blob/master/doc/Strategy.txt which is being used by steam and other things implementing libcapsule. getting the host graphics stack into the container is done through probing the ldcache

<@jsteffan:fedora.im>

16:31:18

i'm still learning, so forgive me if i've misunderstood some of this

<@conan_kudo:matrix.org>

16:34:46

this is... fine

<@conan_kudo:matrix.org>

16:34:52

we're working with the constraints of what we have

<@conan_kudo:matrix.org>

16:35:05

I wish the vulkan people did a better job here though :(

<@conan_kudo:matrix.org>

16:35:14

(ie Khronos)

<@conan_kudo:matrix.org>

16:38:46

daMaestro: for RFC language using "not", the "not" also needs to be treated the same as MUST/SHOULD/MAY/etc.

<@jsteffan:fedora.im>

16:39:10

okay, so it would be `MUST NOT`?

<@conan_kudo:matrix.org>

16:41:05

yes

<@jsteffan:fedora.im>

16:41:15

k, https://datatracker.ietf.org/doc/html/rfc2119#section-2 i'll make those updates now

<@jsteffan:fedora.im>

16:42:34

okay, that is fixed

<@james:fedora.im>

16:43:39

I added a comment. Hopefully we can get +5 in the comments and merge it soon.

<@jsteffan:fedora.im>

16:44:22

per last meeting i was able to attend, i started implementing this in the package set i've been working on for XR and so far haven't ran into any issues -- so that's good at least

<@conan_kudo:matrix.org>

16:45:08

I put a +1 in the ticket

<@conan_kudo:matrix.org>

16:45:21

just asked for the commits to be squashed so that we have a single clean commit

<@jsteffan:fedora.im>

16:46:14

it should be squashed already, so i'll be sure that is the case during the merge

<@james:fedora.im>

16:46:18

Fair enough, I'm mostly against squashes ... but not enough to argue about it for this.

<@conan_kudo:matrix.org>

16:48:06

two commits that say "update guidelines" are kind of pointless

<@conan_kudo:matrix.org>

16:48:30

if they are meaningful commits, for sure I agree

<@conan_kudo:matrix.org>

16:48:37

but stub and fixup commits, less so

<@conan_kudo:matrix.org>

16:50:06

anyway, looks good now from what I see, so +1

<@conan_kudo:matrix.org>

16:50:23

can we get 4 other +1 in the PR?

<@jsteffan:fedora.im>

16:50:56

i oops, it's fixed now.

<@limb:fedora.im>

16:51:05

I added mine

<@james:fedora.im>

16:55:45

Okay, I think we've done as much as we can on that (and looks hopeful to get merged). With the 5 minutes we have left...

<@james:fedora.im>

16:55:47

!link https://pagure.io/packaging-committee/pull-request/1374

<@james:fedora.im>

16:56:04

I think this is a simple merge, but thought I'd ping everyone to see if anybody objects

<@james:fedora.im>

16:57:01

And also if anyone has anything else we can talk about in 4 minutes ... speak up now :)

<@conan_kudo:matrix.org>

16:57:03

this seems fine

<@james:fedora.im>

16:57:32

I will be out next week, and the week after is flock (I'll be there). So probably no meeting the next two weeks.

<@conan_kudo:matrix.org>

16:57:40

👍️

<@limb:fedora.im>

16:58:32

I used a baseball metaphor in a work meeting and I think it was sufficiently out of character that I broke a couple people.

<@conan_kudo:matrix.org>

16:58:42

🤣

<@james:fedora.im>

16:59:14

Sportsball ftw ;)

<@james:fedora.im>

17:00:35

!endmeeting