| *** bauzas_ is now known as bauzas | 05:18 | |
| *** bauzas_ is now known as bauzas | 05:33 | |
| opendevreview | Merged openstack/releases master: Sort output for tools/governance_consistency.py https://review.opendev.org/c/openstack/releases/+/922466 | 08:40 |
|---|---|---|
| *** bauzas_ is now known as bauzas | 08:57 | |
| *** whoami-rajat_ is now known as whoami-rajat | 11:25 | |
| elodilles | reminder: weekly meeting starts in ~30 mins | 12:30 |
| elodilles | #startmeeting releaseteam | 13:00 |
| opendevmeet | Meeting started Fri Jun 28 13:00:10 2024 UTC and is due to finish in 60 minutes. The chair is elodilles. Information about MeetBot at http://wiki.debian.org/MeetBot. | 13:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 13:00 |
| opendevmeet | The meeting name has been set to 'releaseteam' | 13:00 |
| hberaud | o/ | 13:00 |
| elodilles | Ping list: release-team elod | 13:00 |
| elodilles | #link https://etherpad.opendev.org/p/dalmatian-relmgt-tracking | 13:00 |
| frickler | \o | 13:00 |
| elodilles | o/ | 13:00 |
| elodilles | we are down @ line 193 | 13:01 |
| elodilles | R-14 | 13:01 |
| elodilles | let's get started ~o~ | 13:01 |
| elodilles | #topic Review task completion | 13:01 |
| elodilles | 1st & only task was: | 13:01 |
| elodilles | 'Ahead of MembershipFreeze, run governance_consistency.py (frickler)' | 13:02 |
| elodilles | frickler: anything to report? | 13:02 |
| frickler | I ran that last week and showed the results. didn't get to follow up yet | 13:02 |
| frickler | I'll add the list to the tracking pad later | 13:02 |
| elodilles | ACK, thanks | 13:03 |
| frickler | #link https://paste.opendev.org/show/b6cpD8zG0JwYw9w2bhyg/ | 13:03 |
| frickler | that's the list | 13:03 |
| elodilles | +1 | 13:03 |
| elodilles | as I see & as i remember we discussed, that there are only some legitimate deliverables, but those are there for cycles over cycles | 13:04 |
| elodilles | and I don't see anything 'new' | 13:04 |
| elodilles | maybe openapi, codegenerator? | 13:04 |
| elodilles | anyway, we should check with the teams if they want these to be included in 2024.2 Dalmatian | 13:05 |
| frickler | yes, the latter two are new and not ready yet | 13:05 |
| elodilles | ah, so you already pinged the teams, right? | 13:06 |
| elodilles | the ' | 13:06 |
| elodilles | Defined in deliverable files but not in (active) governance' list is also interesting | 13:06 |
| elodilles | i can generate a patch to remove the listed puppet-* repos from deliverables/dalmatian/ | 13:07 |
| frickler | I pinged the cinder team and it seems it is a known issue, need to check the others | 13:07 |
| frickler | the puppet things got retired but not deleted properly due to bad timing I think | 13:08 |
| elodilles | yeah, i think so, too | 13:08 |
| elodilles | btw, i don't find murano-pkg-check under deliverables/dalmatian/ O.o | 13:09 |
| elodilles | how is that possible? :-o | 13:10 |
| elodilles | anyway, as I said, i can propose a patch for the latter listed deliverables | 13:11 |
| frickler | it is independent | 13:12 |
| elodilles | ah, i see | 13:12 |
| elodilles | thanks | 13:12 |
| elodilles | frickler: could you continue to check with the teams for the remaining deliverables in the upper list? | 13:13 |
| elodilles | just to sort these out as soon as possible, so that to have a good view about the Dalmatian deliverables | 13:14 |
| frickler | will do, ack | 13:14 |
| elodilles | thanks o/ | 13:14 |
| elodilles | anything else to add to this task? | 13:14 |
| frickler | I had a small patch to make the output sorted and thus repeatable, that got merged earlier | 13:15 |
| elodilles | ++ | 13:15 |
| frickler | #link https://review.opendev.org/c/openstack/releases/+/922466 | 13:15 |
| elodilles | thanks for that too o/ | 13:15 |
| elodilles | okay, let's move on to the next topic | 13:16 |
| elodilles | #topic Assign R-13 week tasks | 13:16 |
| elodilles | hberaud: would you chair the meeting next week? | 13:16 |
| elodilles | ttx will be on PTO | 13:17 |
| hberaud | no sorry, I'm not fully available next, I leave my place to someone more active than me | 13:17 |
| elodilles | ACK, thanks for the heads up | 13:18 |
| elodilles | then i'm the only candidate, so I'll chair o:) | 13:18 |
| hberaud | not on PTO but with many other topics on my plate | 13:18 |
| elodilles | hberaud: i see :S | 13:18 |
| hberaud | thanks elodilles | 13:19 |
| elodilles | np | 13:19 |
| elodilles | then i added my name to the tasks, feel free to grab any of them if you would like to work on any of them | 13:19 |
| elodilles | move on then | 13:19 |
| elodilles | #topic Review countdown email for week R-13 | 13:20 |
| elodilles | #link https://etherpad.opendev.org/p/relmgmt-weekly-emails | 13:20 |
| elodilles | please review ^^^ | 13:20 |
| ttx | oops | 13:20 |
| ttx | I totally missed the ping | 13:20 |
| ttx | Email LGTM | 13:21 |
| elodilles | np :) we are more or less at a quiet summer period :D | 13:21 |
| frickler | do the last two sections about new deliverables actually apply to anything? | 13:21 |
| frickler | otherwise we might rather want to drop them? | 13:22 |
| hberaud | LGTM | 13:22 |
| elodilles | frickler: yepp, i think yes, we don't have (and probably won't have) new deliverables, so I'm OK with dropping that part | 13:22 |
| elodilles | i've removed it | 13:23 |
| ttx | frickler: i think the remidner can't hurt, even if we do not expect deliverables | 13:23 |
| ttx | also fine removing it | 13:23 |
| elodilles | the reminder is still there about the 'content freeze' | 13:23 |
| ttx | ah perfect | 13:24 |
| elodilles | i just dropped the part about the 'new deliverables' | 13:24 |
| frickler | ack, that's fine then | 13:24 |
| elodilles | good, thanks all for the reviews, will send the mail after the meeting some time | 13:24 |
| hberaud | thanks elodilles | 13:24 |
| elodilles | #topic Open Discussion | 13:25 |
| elodilles | anything to mention? | 13:25 |
| hberaud | yes | 13:25 |
| elodilles | hear hear | 13:25 |
| hberaud | I currently think about stepping down as release manager, as I said few minutes ago, my plate is full of other topics and I'm not enough availabale for the release team | 13:26 |
| elodilles | that sounds a sad news :/ | 13:27 |
| hberaud | I wanted to be transparent with you about my thinking | 13:27 |
| ttx | we do have low expectations! | 13:27 |
| elodilles | as we are a bit small team | 13:27 |
| ttx | I'm also not super active but it all adds up | 13:27 |
| elodilles | yepp, as ttx said, any small help is still a help | 13:27 |
| hberaud | I can remains available to help but with a lower activity | 13:28 |
| frickler | that would be much better than stepping down completely IMO, too | 13:28 |
| elodilles | hberaud: we appreaciate any help, so that's fine by us | 13:28 |
| hberaud | ok, then, will do my best | 13:28 |
| elodilles | thanks in advance! | 13:29 |
| elodilles | and thanks for your work so far o/ | 13:29 |
| ttx | That's fine by me, thanks for setting expectations correctly ! Just doing a review day is already moving the needle (that's the minimum I try to force myself to | 13:29 |
| ttx | cover) | 13:29 |
| frickler | feel free to ignore pings if you don't have time | 13:29 |
| hberaud | thanks all | 13:29 |
| elodilles | yepp, as frickler says ++ | 13:29 |
| hberaud | I'll be happy to reply to pings if I can | 13:30 |
| elodilles | +1 | 13:30 |
| hberaud | that's all for me | 13:30 |
| frickler | but also, if you decide you'd rather quit completly, don't listen to us | 13:30 |
| hberaud | frickler: sure :) | 13:31 |
| hberaud | I can still do the minimal service | 13:31 |
| elodilles | o7 | 13:31 |
| elodilles | any other topic or merrier news? o:) | 13:33 |
| hberaud | nope | 13:34 |
| elodilles | if not, then thanks all for participation and see you next week o/ | 13:34 |
| elodilles | #endmeeting | 13:34 |
| opendevmeet | Meeting ended Fri Jun 28 13:34:39 2024 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 13:34 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/releaseteam/2024/releaseteam.2024-06-28-13.00.html | 13:34 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/releaseteam/2024/releaseteam.2024-06-28-13.00.txt | 13:34 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/releaseteam/2024/releaseteam.2024-06-28-13.00.log.html | 13:34 |
| frickler | o/ | 13:34 |
| hberaud | thx elodilles | 13:34 |
| elodilles | o/ | 13:37 |
| ttx | thanks elodilles ! Sorry for being late :) | 13:37 |
| opendevreview | Elod Illes proposed openstack/releases master: Remove inactive deliverables from dalmatian https://review.opendev.org/c/openstack/releases/+/923052 | 13:45 |
| elodilles | ttx: i would say, that 'no problem, you can be here in time next week', but as you signalled you'll be on PTO, so enjoy your free time and the nice hot summer :) o/ | 13:48 |
| opendevreview | Elod Illes proposed openstack/releases master: Remove python-muranoclient from Dalmatian https://review.opendev.org/c/openstack/releases/+/923053 | 13:54 |
| opendevreview | Elod Illes proposed openstack/releases master: Mark murano-pkg-check as retired https://review.opendev.org/c/openstack/releases/+/923054 | 14:19 |
| *** bauzas_ is now known as bauzas | 17:17 | |
| dhellmann | @fungi hey! I saw your post on the python discussion forum. You might find https://github.com/python-wheel-build/fromager interesting, if not actually useful. | 19:55 |
| fungi | dhellmann: thanks! and yeah, i saw another post link to that not long ago | 19:56 |
| dhellmann | it's a new thing, so I'm doing a bit of link sharing for awareness | 19:56 |
| fungi | i think for us the trick is to find out exactly which packages pip install would install transitively and then build wheels for the ones which only had sdists matching the requirement | 19:57 |
| dhellmann | some changes it might need to be fully useful for you would be to understand constraints files (it works with requirements files now) and to have a flag to say "if you find a wheel, just use it" (that's a per-package setting right now) | 19:57 |
| dhellmann | yep, fromager is trying to build everything by default right now | 19:57 |
| fungi | luckily, constraints files the way we're using them are essentially also functional requirements files | 19:57 |
| fungi | so you could pip install -r the constraints file instead of passing it to -c anyway | 19:58 |
| dhellmann | true | 19:58 |
| fungi | syntax in the files turns out to be compatible | 19:58 |
| dhellmann | that was on purpose, iirc | 19:59 |
| fungi | one of the reasons lifeless's original design for pip's constraints feature was so good ;) | 19:59 |
| * dhellmann nods | 19:59 | |
| fungi | just too bad it took the pip maintainers ignoring his dependency resolver work and then redoing it a decade later | 20:00 |
| fungi | to get us a real dep solver | 20:00 |
| fungi | so much of what we were doing then was just too far ahead of the needs of the broader python packaging community, i think | 20:01 |
| dhellmann | yeah, exactly | 20:01 |
| fungi | so stuff like that got rejected because "nobody will ever need this/what we have is good enough" so i'm thrilled they did actually take the constraints feature | 20:01 |
| dhellmann | that's part of why I'm trying to limit the complexity of build variant stuff | 20:02 |
| dhellmann | yeah, that constraints file feature is a lifesaver | 20:02 |
| fungi | while the variants stuff probably isn't something we'll need, i've been following the discussion around it with great interest nonetheless | 20:03 |
| fungi | but yeah, fromager looks like it could be a drop-in replacement for what we're doing with pip in wheel cache jobs | 20:03 |
| clarkb | the opendev build/base python container images do something similar too | 20:03 |
| dhellmann | these AI packages are _massive_ | 20:03 |
| clarkb | but the use case is far more constrained (basically build wheels for this specific platform, then install those wheels into the target container) | 20:03 |
| dhellmann | I'd be happy to entertain feature patches if it's missing something you need | 20:03 |
| fungi | the alternative we're looking at is just not doing it any more, enough projects finally publish wheels that our corner cases are alternative processors (aarch64 testing) or trying to use newer python interpreters with packages that pre-date them | 20:04 |
| dhellmann | less may be more, yeah | 20:04 |
| fungi | i think libvirt-python is the main sticking point for us, it's sdist-only still | 20:05 |
| dhellmann | if it's just 1 or 2 packages, managing only those might be easier than managing all of the automation? OTOH, if what you have is working... | 20:06 |
| fungi | yeah, that's the pivot point we're thinking we might be near/at | 20:07 |
| fungi | there's a sweet-spot between the work to keep the existing implementation maintained and the work to get along without it | 20:07 |
| fungi | dhellmann: looking at the fromager overview, the main thing i think it might not do is reuse "good enough" wheels (e.g. manylinux/abi3) from pypi and only build wheels for things that aren't installable on the target platform? | 20:09 |
| fungi | what we've been trying to do with our current solution is only cache platform-specific wheels centrally for things that pip would otherwise rebuild from sdist on its own | 20:10 |
| fungi | so that we reuse available wheels on pypi whenever possible | 20:10 |
| clarkb | one important reason for that is people running your software in the real world will get the pypi stuff | 20:11 |
| clarkb | so you don't want to overoptimize and paper over problems you might get in the real world | 20:11 |
| fungi | yeah, we're basically trying to emulate a generic `pip install ...` as closely as possible, just accelerate invocation by not requiring rebuilds from source in the specific cases where it would do that | 20:12 |
| fungi | we specifically want to avoid rebuilding everything from sdist | 20:12 |
| fungi | and not so much because of the resource utilization, but because of what clarkb mentions about reusing what's on pypi since that's what users will also be installing | 20:13 |
| fungi | also, notably, our solution is intentionally pip-specific since we're concerned with what `pip install` will decide to do and optimizing for that case | 20:14 |
| fungi | swapping pip out for uv or whatever is out of scope in our case | 20:15 |
| dhellmann | fungi : yeah, I could see adding a flag to bootstrap to say it should default all packages to pre-built if it can | 20:15 |
| dhellmann | and yeah, we're using pip to build container images out of the wheels that fromager builds, but we _do_ want the special optimized builds | 20:16 |
| fungi | i can definitely see the case for preferring optimized rebuilds, it's just not our focus since we're currently more concerned with correctness and testing as closely as possible to what users are going to end up installing | 20:20 |
| fungi | so a preference for "bug-compatible" behavior in most cases | 20:21 |
| fungi | but yeah, overall fromager looks really cool, thanks again for pointing it out! | 20:22 |
| *** bauzas_ is now known as bauzas | 20:55 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!