13:00:34 #startmeeting NeuroFedora - 2023-06-19 13:00:34 Meeting started Mon Jun 19 13:00:34 2023 UTC. 13:00:34 This meeting is logged and archived in a public location. 13:00:34 The chair is FranciscoD_. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions. 13:00:34 Useful Commands: #action #agreed #halp #info #idea #link #topic. 13:00:34 The meeting name has been set to 'neurofedora_-_2023-06-19' 13:00:39 #meetingname neurofedora 13:00:39 The meeting name has been set to 'neurofedora' 13:00:49 #info Agenda is on the blog: https://neuroblog.fedoraproject.org/2023/06/19/next-open-neurofedora-meeting-19-June-1300-utc.html 13:00:56 #info bot commands: https://fedoraproject.org/wiki/Meeting:Guide#MeetBot_Commands 13:01:01 #topic Introductions and roll call 13:01:06 .hellowmynameis ankursinha 13:01:44 .hellomynameis ankursinha 13:01:45 FranciscoD_: ankursinha 'Ankur Sinha' 13:01:53 there we go 13:02:11 I'll wait here till 5 past 13:06:11 .members neuro-sig 13:06:13 FranciscoD_: Members of neuro-sig: sagitter, ignatenkobrain, bt0dotninja, zbyszek, sergiopr, blackfile, ankursinha, lbazan, ilgrad, gicmo, alciregi, nerdsville, mhough, music, major, fangq, anilbey, iztokf, aekoroglu, gui1ty, shaneallcroft, hardeborlaa, mairacanal, vanessakris 13:06:44 #topic Tasks from last meeting 13:06:52 #info last meeting logs are here: https://meetbot.fedoraproject.org/latest/neurofedora 13:07:23 #info FranciscoD_ open issue/PR in package maintainer docs about re-triggering CI on PRs: PENDING 13:07:29 forgot about these, re-assigning 13:07:33 #action FranciscoD_ open issue/PR in package maintainer docs about re-triggering CI on PRs 13:07:47 #info FranciscoD_ start non-responsive maintainer checks for Ilya: PENDING 13:07:51 #action FranciscoD_ start non-responsive maintainer checks for Ilya 13:08:18 #info Penguinpee ask around if Koschei can be made to send notifications to the ML or automatically file bugs: PENDING (since Penguinpee has been on vacation since last meeting from what I know) 13:08:19 #action Penguinpee ask around if Koschei can be made to send notifications to the ML or automatically file bugs 13:08:31 #topic Open pagure tickets 13:08:35 #info https://pagure.io/neuro-sig/NeuroFedora/issues?status=Open&tags=S%3A+Next+meeting 13:09:08 #info https://pagure.io/neuro-sig/NeuroFedora/issue/549 -> NeuroFedora at Flock: we've submitted a CFP 13:10:47 #info https://pagure.io/neuro-sig/NeuroFedora/issue/551 -> Discussion on EPEL support: to be discussed in the ticket, please do take a look and comment 13:10:54 .hello gui1ty 13:10:55 Penguinpee: gui1ty 'Sandro .' 13:11:06 Sorry for being late. Got sidetracked. 13:11:43 Yeah, I'll pick up the Koschei notifications issue again. 13:13:06 ah, no worries, I thought you were still on vacation? 13:13:06 although I did see your replies on the -devel list today, which made me question my thought XD 13:13:20 #info https://pagure.io/neuro-sig/NeuroFedora/issue/552 -> using packit to help with neuro-sig package maintenance: only opened this a few hours ago, so we'll discuss it next meeting when folks have had a chance to look into things 13:14:02 Packit looks promising. I may try it on some of my own packages in the course of the coming weeks. 13:15:10 I got back from vacation Thursday night. Back to work tomorrow. It took me the better part of the weekend to catch up. 13:15:46 Cool, I took over the vxl ticket in the meantime, have made some headway 13:16:15 yeh, packit looks very useful, if it can help us with some of the simpler package update type tasks, that'll help us 13:16:28 Nice! I'll leave vxl to you then. 13:16:37 Sure thing. I haven't been doing any builds in the past week because I'm waiting for the python3.12 builds to complete 13:17:26 Yeah. I saw that. I still have other, non-Python, packages I need to attend to. 13:17:42 #info https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/message/ZTID6MR7TU2X6ZKDGGH7RR426ODUEENX/ -> python 3.12 rebuild is in progress, please hold off building python and related packages until this is complete 13:18:19 # 13:18:33 #info Python 3.12 side tag on Koji: https://koji.fedoraproject.org/koji/builds?latest=0&tagID=f39-python&order=-build_id&inherited=0 13:18:49 #topic Packages health check 13:19:19 #info Neuro-sig packager dashboard: https://packager-dashboard.fedoraproject.org/dashboard?groups=neuro-sig 13:19:41 #info Please ignore s390x FTBFS reports if the package exludes s390x as a build arch. 13:20:17 #info A few python packages have new versions available: we'll wait until the py3.12 rebuild is over to tackle these 13:20:51 #info Same applies to ftbfs bugs 13:21:31 Well, we can test the FTBFS packages on the stable branches, can we not? 13:22:08 yeh, stable branches are good 13:22:26 #info Continue FTBFS investigations on stable branches (since they're unaffected by the re-build) 13:23:04 packages going FTBFS on stable branches usually indicate a dep update, which means the maintainers of packages did not do a complete impact assessment---this was being discussed in the -devel thread 13:25:01 I hope to be able to spare some cycles for working on FTBFS packages. New releases could be queued by way of PR, which we only merge after the Python 3.12 side tag has been merged. 13:27:59 I just looked at the FTBFS (F38) for python-neurotune: No dependency changes. Might be a false positive. There has been some discussion on the devel list about noarch packages failing on some archs (e.g. i686). 13:28:41 #info FTBFS reports for these packages in stable branches: arbor (f37), genesis-simulator (f38), getdp (f38), highfive (f38), libneurosim (f38), libsbml (f38), libsonata (f37), morphio (f38), python-astor (f37), python-bluepyopt (f37, f38), python-cyiopt (f37), python-dipy (f38), python-mne (f37, f38), python-neatdend (f37), python-neurotune (f38), python-nibabel (f37), python-nilearn (f38), python-nipy (f38, f37), python-pandas (f37), 13:28:41 python-pyswarms (f38, f37), python-pyunicorn (f38, f37), python-steps (f38) 13:29:39 Yeh -> https://koschei.fedoraproject.org/package/python-neurotune?collection=f38 13:29:56 OSError: [Errno 28] No space left on device: '/tmp/mock-resolv.dbitjql0' 13:30:21 yeh, these will fix themselves when koschei checks them next 13:30:29 Would a simple scratch build make this go away on the Koschei dashboard? 13:30:39 #info Some FTBFS koschei reports could be false positives -> builders ran out of space 13:31:07 Penguinpee: I don't think so, but I don't know. I think koschei has to do the build. No harm trying? 13:31:33 Sure. I'll try. 13:32:54 the wiki page doesn't say anything about manual re-triggers: https://fedoraproject.org/wiki/Koschei#Your_packages 13:35:54 Other than the FTBFS issues, packages are fine 13:36:06 #topic Open package reviews check: https://bugzilla.redhat.com/show_bug.cgi?id=fedora-neuro 13:36:11 #undo 13:36:11 Removing item from minutes: 13:36:19 #topic Open package reviews check 13:36:28 #info Neuro-sig reviews tracker bug: https://bugzilla.redhat.com/show_bug.cgi?id=fedora-neuro 13:36:48 #info Please remember to block this when submitting a new review so that all sig members are notified 13:37:11 Oh. I didn't know about the wiki. Thanks for the link. I may have spotted the answer regarding FTBFS bugs. But I'll ask around nonetheless. 13:38:16 #info one review ticket open: https://bugzilla.redhat.com/show_bug.cgi?id=2213973 -> should be easy, if someone has the cycles, please take it up (blocks the python-bokeh update) 13:38:45 How are new packages affected by the mass rebuild for Python 3.12? 13:39:20 hrm, I think the review can go ahead, but one should perhaps not commit/build until the rebuild is completed 13:39:58 #action FranciscoD_ mail -devel about review swap if package not picked up in a few days 13:40:12 Or notify the Pythonistas doing the mass rebuild? 13:41:31 Yeh, could do---but they'll have to update their package list and then run the build etc. 13:41:47 probably simpler to wait for the mass rebuild to finish and then add new packages 13:44:15 #topic Comp neuro image compose check 13:44:20 #info koji task here: https://koji.fedoraproject.org/koji/packageinfo?packageID=30691 13:44:34 #info Green tick, all going along fine here 13:44:53 #topic Neuro(science) query of the week 13:45:46 #info Not neuro related, but one related to all the LLMs: https://venturebeat.com/ai/the-ai-feedback-loop-researchers-warn-of-model-collapse-as-ai-trains-on-ai-generated-content/ 13:46:13 I don't remember coming across anything else too interesting 13:48:14 #info https://www.sciencenews.org/article/supermassive-black-hole-orbit-blazar 13:48:18 #info https://www.sciencenews.org/article/one-photon-photosynthesis-light 13:48:54 #info https://blog.arxiv.org/2023/06/12/arxiv-is-hiring-software/ (job opportunity at arxiv) 13:49:44 that's all I can find on the planet 13:49:56 #topic next meeting day, chair 13:50:44 #info same time (1300 UTC), in 2 weeks (03 July) 13:50:50 #info ankursinha will chair 13:50:54 #topic Open floor 13:51:29 we've got ~10 minutes here, for anything that wasn't discussed already 13:52:00 I haven't read the full LLM article yet. But the headline makes it sound like other diseases caused by feeding an organism with stuff it wouldn't eat by itself, e.g CJD (aka mad cow disease). 13:54:43 yeh, I guess the thing is---what is the LLM supposed to eat? If it's meant to "eat the whole internet" as it does now, it will end up eating lots of things it and other LLMs produce 13:55:08 the article does point out some fixes, I'm sure the big LLM corps are working on these already 13:55:30 I tried to access your proposal for Flock. But I was nagged into filling out a form and didn't bother. Is it available somewhere else FranciscoD_? 13:55:58 "I'm sure the big LLM corps are working on these" <-- For sure. They need to protect their assets. 13:56:49 Penguinpee: I can update the comment with the text, I think one has to sign into the flock cfp portal to see it there 13:58:16 Yeah. And upon signing in I have to provide details. Some of which I believe should be pre-filled with FAS data if that is the login method one chooses. 13:59:06 No biggie. I'll try again and may nag some people as to why all that is required. 13:59:16 +1 13:59:20 I added the abstract etc. to our ticket 13:59:43 ankursinha++ 13:59:52 FranciscoD_++ 14:00:11 .fasinfo ankursinha 14:00:12 Penguinpee: User: ankursinha, Name: Ankur Sinha, Email: sanjay.ankur@gmail.com, Creation: 2008-05-28T06:16:53, IRC Nicks: FranciscoD and FranciscoD_, Timezone: Europe/London, Locale: en-GB, GPG Key IDs: E629112D, Status: None 14:00:15 Penguinpee: Groups: fedora-contributor, ask-fedora, fedora-join, fedorabugs, freemedia, gitfedora-tour, gitfpaste, neuro-sig, packager, provenpackager, qa, robotics-sig, scitech, scitech_sig, wikiedit 14:00:30 No cookies for you, I guess. ;) 14:00:36 no cookies till the next release cycle :( 14:00:37 yeh 14:01:03 Cool, thanks for coming as usual Penguinpee . We've hit the hour mark, I'll close the meeting now and send out the logs 14:01:05 What does zodbot think? One cannot live on one cokkie per release alone. 14:01:05 #endmeeting