#meeting:fedoraproject.org log

<@ankursinha:fedora.im>

13:02:01

!startmeeting NeuroFedora - 2025-02-10

<@meetbot:fedora.im>

13:02:02

Meeting started at 2025-02-10 13:02:01 UTC

<@meetbot:fedora.im>

13:02:02

The Meeting name is 'NeuroFedora - 2025-02-10'

<@ankursinha:fedora.im>

13:02:04

!meetingname neurofedora

<@meetbot:fedora.im>

13:02:05

The Meeting Name is now neurofedora

<@ankursinha:fedora.im>

13:02:14

!info Agenda: https://neuroblog.fedoraproject.org/2025/02/10/next-open-neurofedora-meeting-10-february-2025-1300-utc.html

<@ankursinha:fedora.im>

13:02:18

!topic Intros/roll call

<@ankursinha:fedora.im>

13:02:20

!hi

<@zodbot:fedora.im>

13:02:22

Ankur Sinha (ankursinha) - he / him / his

<@music:fedora.im>

13:03:22

!hello

<@zodbot:fedora.im>

13:03:27

Benjamin Beasley (music) - he / him / his

<@ankursinha:fedora.im>

13:04:10

!group members neuro-sig

<@zodbot:fedora.im>

13:04:13

Members of neuro-sig: Ali Erdinc Koroglu, alciregi, Ankur Sinha, Alberto Rodriguez Sanchez, fangq, Sandro ., Igor Raits, ilgrad, iztokf, lbazan, Aniket Pradhan, mhough, Benjamin Beasley, sagitter, sergiopr, Zbigniew Jędrzejewski-Szmek

<@ankursinha:fedora.im>

13:04:33

Alessio: Luis Bazan ^

<@ankursinha:fedora.im>

13:04:55

Everyone else is welcome to join us too, of course :)

<@ankursinha:fedora.im>

13:05:40

!topic Tasks from last meeting

<@ankursinha:fedora.im>

13:05:55

!info Last meeting logs are here: https://meetbot.fedoraproject.org/meeting_matrix_fedoraproject-org/2025-01-27/neurofedora.2025-01-27-13.01.html

<@ankursinha:fedora.im>

13:06:11

I have a few pending tasks that are all WIP. The one that I'm currently working on is the dcmtk update

<@ankursinha:fedora.im>

13:06:25

!info dcmtk update in progress here: https://src.fedoraproject.org/rpms/dcmtk/pull-request/10

<@ankursinha:fedora.im>

13:06:54

the new versions fix a couple of CVEs too, so if the impact check goes well, I'll consider updating F40+

<@gui1ty:fedora.im>

13:07:02

!hi

<@zodbot:fedora.im>

13:07:03

Sandro . (gui1ty)

<@ankursinha:fedora.im>

13:07:29

Penguinpee: music: does that sound fine, do you think? I need to run an abidiff check first, of course

<@ankursinha:fedora.im>

13:07:59

impact check is running here on COPR, by the way: https://copr.fedorainfracloud.org/coprs/ankursinha/dcmtk-3.6.9/builds/

<@ankursinha:fedora.im>

13:08:04

itk is still building

<@music:fedora.im>

13:08:29

You said in the PR that it bumps the SONAME version, so it would have to be considered ABI-incompatible and dependent packages would have to be rebuilt.

<@ankursinha:fedora.im>

13:08:52

Yeh, they bump the patch number for each release, and the patch number goes into the soname version

<@music:fedora.im>

13:08:53

I left some notes on the process for shipping an SONAME bump in stable releases: https://src.fedoraproject.org/rpms/dcmtk/pull-request/10

<@ankursinha:fedora.im>

13:09:18

So it's not a clear indicator of ABI changes. We'll have to run abidiff etc. to get details

<@ankursinha:fedora.im>

13:09:26

Ah, cool, I'll take a look at your notes

<@music:fedora.im>

13:10:06

Even if the ABI is otherwise compatible, if the SONAME version has changed, it’s still an incompatible update because all dependent packages must be rebuilt.

<@ankursinha:fedora.im>

13:10:24

Yes, that's what I'm currently testing on COPR

<@music:fedora.im>

13:11:01

If it’s like most C++ libraries, the rules for maintaining ABI stability are so Byzantine that upstream doesn’t see it worth it to attempt to guarantee anything.

<@tslee_:matrix.org>

13:11:17

!hi Lee from India

<@zodbot:fedora.im>

13:11:20

Sorry, I can only look up one username at a time

<@music:fedora.im>

13:13:07

…and if you have to rebuild all dependent packages due to an SONAME version bump, then you’re no longer concerned about ABI diffing, only API compatibility.

<@ankursinha:fedora.im>

13:13:13

Hi Lee! Welcome!

<@ankursinha:fedora.im>

13:13:13

the other option is to only backport the specific commits that fix the CVEs to to F40/F41---no soname bump if we do that

<@ankursinha:fedora.im>

13:13:14

I'll see what works once the impact check is complete and I've run abipkgdiff etc.

<@ankursinha:fedora.im>

13:13:41

music: do you know, off the top of your head, the command to run fedabipkgdiff for this sort of thing?

<@music:fedora.im>

13:13:49

Yes, that is probably the better way to go if they are self-contained enough to backport and don’t inherently break the ABI.

<@ankursinha:fedora.im>

13:14:32

I guess once rawhide is built and up to date, I can run `fedabipkgdiff --from fc42 --to fc43` and that should do it

<@ankursinha:fedora.im>

13:14:46

(Probably something that should be documented in the package maintainer docs)

<@music:fedora.im>

13:15:15

Yes, I do… hold on, let me type it out.

<@ankursinha:fedora.im>

13:15:24

Lee: we just go through the agenda here to check on our packages and tasks, so please chime in with any comments at all as you wish :)

<@tslee_:matrix.org>

13:15:27

Thanks

<@tslee_:matrix.org>

13:16:18

noted

<@ankursinha:fedora.im>

13:16:34

!info dcmtk update is in progress: needs impact check + ABI change check for us to decide what releases it should land in and further steps

<@music:fedora.im>

13:16:48

So it’s `fedabipkgdiff --from fc43 ./path/to/package-or-subpackage-with-shared-libraries.fc43.x86_64.rpm`, but there’s a quirk.

<@music:fedora.im>

13:17:20

It will find the `debugsource` package, get confused, and bail out with `Found more than one debuginfo package in this directory. At the moment, fedabipkgdiff is not able to deal with this case. Please create two separate directories and put an RPM and its ancillary debuginfo and devel RPMs in each directory.`

<@ankursinha:fedora.im>

13:17:34

AH, I did get that error!!

<@ankursinha:fedora.im>

13:17:38

and then I gave up XD

<@music:fedora.im>

13:17:52

So what I do is make an empty directory and copy just one (sub)package containing shared libraries into it, along with the accompanying `debuginfo` package. Then it works.

<@music:fedora.im>

13:18:31

For packages that ship different shared libraries in different subpackages (not dcmtk), rinse and repeat for each subpackage with shared libraries.

<@ankursinha:fedora.im>

13:18:40

Cool

<@ankursinha:fedora.im>

13:19:25

!info `fedabipkgdiff` usage: `fedabipkgdiff --from fc43 ./path/to/package-or-subpackage-with-shared-libraries.fc43.x86_64.rpm` but it needs each sub-package containing shared libraries and the accompanying debuginfo package to be moved to a new clean directory.

<@music:fedora.im>

13:20:33

Again, if there is already an SONAME version bump you don’t really care much about ABI changes in Rawhide or F42, but fedabipkgdiff is useful for double-checking that a backported patch didn’t break ABI.

<@ankursinha:fedora.im>

13:20:34

!info Ankur remove his name from astro + scientific kickstart -> DONE (Ankur wasn't listed in these, so nothing needed here)

<@ankursinha:fedora.im>

13:21:09

!action Ankur Ankur re-check bokeh PR and update for new versions

<@ankursinha:fedora.im>

13:21:36

this one requires some work to re-do the license etc. check with the new version. My PR there is out of date (by a year now!)

<@ankursinha:fedora.im>

13:21:59

Penguinpee: look at exdir and plotly

<@ankursinha:fedora.im>

13:22:09

Penguinpee: do you remember what this was about? ^

<@ankursinha:fedora.im>

13:22:29

!action Ankur open PR porting over comp-neuro image to Kiwi

<@ankursinha:fedora.im>

13:22:33

I've begun, but not finished

<@gui1ty:fedora.im>

13:22:35

No, I don't. I just had a quick search, but nothing popped up.

<@gui1ty:fedora.im>

13:23:41

I'll try to retrace to figure out. But it may well be that this is no longer relevant.

<@ankursinha:fedora.im>

13:26:13

!info Ticket for porting to Kiwi: https://pagure.io/neuro-sig/NeuroFedora/issue/577

<@ankursinha:fedora.im>

13:26:28

I haven't assigned it to myself yet, in case others want to have a go. When I do work on it properly, I'll take it up

<@ankursinha:fedora.im>

13:28:31

!topic Open Pagure tickets

<@ankursinha:fedora.im>

13:28:46

!info Tickets are here: https://pagure.io/neuro-sig/NeuroFedora/issues?status=Open&tags=S%3A+Next+meeting

<@ankursinha:fedora.im>

13:28:50

!info No new tickets

<@ankursinha:fedora.im>

13:29:22

Wrt STEPS: I have it building. The updated file lists etc. need to be looked at next

<@ankursinha:fedora.im>

13:29:28

!info STEPS update: https://pagure.io/neuro-sig/NeuroFedora/issue/566

<@ankursinha:fedora.im>

13:31:00

!topic Package health check

<@ankursinha:fedora.im>

13:31:18

!info Neuro-SIG packages on the dashboard are here: https://packager-dashboard.fedoraproject.org/dashboard?groups=neuro-sig

<@ankursinha:fedora.im>

13:33:57

music: I see you fixed iv, did that no magically fix neuron by any chance? :P

<@gui1ty:fedora.im>

13:35:16

It doesn't look like it according to Koschei.

<@ankursinha:fedora.im>

13:35:21

I'm just firing off a quick scratch build to check too

<@ankursinha:fedora.im>

13:35:41

That would've been lucky :)

<@gui1ty:fedora.im>

13:36:02

https://koschei.fedoraproject.org/build/19617468

<@gui1ty:fedora.im>

13:36:02

Assuming this has the fixed `iv` you referred to.

<@ankursinha:fedora.im>

13:38:39

hrm, too bad. We'll look into it then

<@ankursinha:fedora.im>

13:38:53

I think we can let mod2c go, given that upstream has archived the repo: https://bugzilla.redhat.com/show_bug.cgi?id=2340871

<@ankursinha:fedora.im>

13:39:43

I can retire it in F42+ now

<@gui1ty:fedora.im>

13:40:14

Now is the time...

<@gui1ty:fedora.im>

13:40:27

It's a leave package as well.

<@ankursinha:fedora.im>

13:40:40

yeh

<@ankursinha:fedora.im>

13:40:59

I guess I should remove neuro-sig from it first before I retire it

<@gui1ty:fedora.im>

13:42:02

Good idea.

<@ankursinha:fedora.im>

13:43:37

Does retiring not also orphan the package?

<@ankursinha:fedora.im>

13:43:40

doesn't seem to

<@ankursinha:fedora.im>

13:45:31

I'll prioritise the neuron FTBFS because that'll fix a lot of others that depend on it

<@ankursinha:fedora.im>

13:45:56

!action Ankur prioritise looking at neuron FTBFS: https://bugzilla.redhat.com/show_bug.cgi?id=2340926

<@gui1ty:fedora.im>

13:46:13

I'm not exactly sure how those processes work behind the scenes, but I think it should be one or the other. Orphaning entails retirement unless someone adopts the package.

<@ankursinha:fedora.im>

13:47:32

I retired it, and now have also orphaned it

<@ankursinha:fedora.im>

13:47:52

(since it's inactive upstream, I wouldn't expect anyone to take it on)

<@gui1ty:fedora.im>

13:48:22

Let's what gremlins you've been feeding...

<@gui1ty:fedora.im>

13:48:31

Let's see what gremlins you've been feeding...

<@ankursinha:fedora.im>

13:48:39

Let's move on to the next topic, we're short on time now :o

<@ankursinha:fedora.im>

13:48:45

!topic Open package reviews check

<@ankursinha:fedora.im>

13:49:12

!info Neuro-sig new package review tracker is here: https://bugzilla.redhat.com/show_bug.cgi?id=fedora-neuro

<@ankursinha:fedora.im>

13:50:01

!info Only one review ticket awaiting review at the moment: https://bugzilla.redhat.com/show_bug.cgi?id=2283111 (python-astrovascpy)

<@ankursinha:fedora.im>

13:50:19

!topic CompNeuro lab iso image compose check

<@ankursinha:fedora.im>

13:50:34

!info See the task on koji here: https://koji.fedoraproject.org/koji/packageinfo?packageID=30691

<@ankursinha:fedora.im>

13:50:38

it's currently broken..

<@ankursinha:fedora.im>

13:51:08

that's because of our FTBFS bugs

<@ankursinha:fedora.im>

13:51:20

!info Compose currently broken until we fix neuron etc FTBFS issues

<@ankursinha:fedora.im>

13:51:37

!topic (neuro)science bits of the week

<@ankursinha:fedora.im>

13:52:05

Well, this sort of thing has been the focus for most academics recently

<@ankursinha:fedora.im>

13:52:10

!info https://www.science.org/content/article/nih-slashes-overhead-payments-research-sparking-outrage

<@ankursinha:fedora.im>

13:52:23

and this

<@ankursinha:fedora.im>

13:52:29

!info https://www.nature.com/articles/d41586-025-00396-6

<@ankursinha:fedora.im>

13:53:05

!info More on "planet neuroscientists" here: https://neuroblog.fedoraproject.org/planet-neuroscientists/

<@ankursinha:fedora.im>

13:53:47

This is interesting

<@ankursinha:fedora.im>

13:53:50

!info https://www.womeninneuroscienceuk.org/post/adhd-the-brain-what-s-in-the-matter

<@ankursinha:fedora.im>

13:57:28

!topic Next meeting

<@ankursinha:fedora.im>

13:57:35

In 2 weeks at the same time I guess

<@ankursinha:fedora.im>

13:57:48

!info Next meeting in 2 weeks on 24th Feb at 1300 UTC

<@ankursinha:fedora.im>

13:57:53

!topic Open floor

<@tslee_:matrix.org>

13:58:04

<@ankursinha:fedora.im>

13:58:05

3 minutes for anything else, and then we can continue in #neuro:fedoraproject.org :)

<@tslee_:matrix.org>

13:58:15

I have a doubt

<@ankursinha:fedora.im>

13:58:42

Sure, how can we help?

<@tslee_:matrix.org>

13:59:07

Where does the memory get stored when we are talking to a person ?

<@ankursinha:fedora.im>

13:59:33

There are multiple answers to that, depending on what level you want to look at

<@tslee_:matrix.org>

13:59:38

In AI context-window is the word I think

<@ankursinha:fedora.im>

13:59:44

Ah, no

<@ankursinha:fedora.im>

13:59:58

the AI context window is a machine learning/AI related term

<@ankursinha:fedora.im>

14:00:14

it's "inspired" by biological memory, but it does not apply to the brain

<@ankursinha:fedora.im>

14:00:28

in the brain, memory is stored in synapses between neurons

<@ankursinha:fedora.im>

14:00:55

Machine learning takes this inspiration from this and so "memory" in machine learning is also stored in "synapses" between neurons

<@tslee_:matrix.org>

14:01:09

I want to know what is the equivalent in Real AI

<@ankursinha:fedora.im>

14:01:31

Lee: should we take this back to #neuro:fedoraproject.org ? Someone else may want to use this room now

<@ankursinha:fedora.im>

14:01:35

!endmeeting