| tonyb | fungi, gouthamr: Thanks. | 00:41 |
|---|---|---|
| opendevreview | Goutham Pacha Ravi proposed openstack/governance master: [resolution] Extend scope of VMT to cover all project teams https://review.opendev.org/c/openstack/governance/+/944817 | 05:19 |
| opendevreview | Goutham Pacha Ravi proposed openstack/governance master: [resolution] Extend scope of VMT to cover all project teams https://review.opendev.org/c/openstack/governance/+/944817 | 05:28 |
| dansmith | fungi: kinda getting a lot of spam to the list l lately, eh? | 13:52 |
| dansmith | I assume these are subscribing and then posting? | 13:52 |
| fungi | dansmith: yes, i saw one over the weekend from a subscriber who i blocked and then deleted the message from the archive. then yesterday a different new subscriber sent two more, so i did the same with them | 13:53 |
| fungi | i haven't seen the list posts this morning yet, working my way through my usual morning backlog, but if there are more i may adjust the list config so any new subscribers get their posts held for moderation and then a moderator can switch them to allowed to post without moderation if the message they send is legit | 13:55 |
| fungi | i'd rather not have to do that since it's more work for the list moderators, but we can't afford to have the ml turned into a spam emitter either | 13:56 |
| dansmith | Yeah I started doing that on my mm3 list, default to moderate, unmoderate when I approve their first message | 13:56 |
| dansmith | yeah, it's a major pain | 13:56 |
| dansmith | but hoping they're going to leave you alone or become better humans is probably not realistic | 13:57 |
| fungi | agreed | 13:57 |
| clarkb | I'm not sure where the uwsgi alternatives discussion for openstack ended up but trying to build opendev's uwsgi container images has been breaking on segfaults during uwsgi's compilation process. This has pushed me to start looking at alternatives for opendev and I've begun experimenting with granian in https://review.opendev.org/c/opendev/system-config/+/944806 and | 15:28 |
| clarkb | https://review.opendev.org/c/opendev/system-config/+/944806 | 15:28 |
| clarkb | reasons I looked at granian: it supports asgi, rsgi (granians own thing), and wsgi. It has prebuilt wheels for x86_64 and aarch64 so no building from scratch. And as a bonus claims to perform well | 15:31 |
| dansmith | how new is it? and is it packaged in distros? | 15:34 |
| dansmith | looks like first non-beta release in late 2022 | 15:36 |
| clarkb | ya and v1 was january 2024 | 15:36 |
| clarkb | doesn't appear to be packaged in ubuntu | 15:37 |
| noonedeadpunk | clarkb: that sounds really interesting | 15:37 |
| dansmith | I guess I'm less interested because I prefer things that are more well-understood and I don't get excited just because something is written in rust :P | 15:38 |
| noonedeadpunk | I wonder what would it take to build wheels locally though | 15:38 |
| noonedeadpunk | I was exzcited about asgi/wsgi/http2/tls mainly | 15:38 |
| dansmith | like I dunno what it would take to get this into RHEL/ubuntu(non-universe) | 15:38 |
| dansmith | uwsgi isn't even in rhel, AFAIK | 15:39 |
| noonedeadpunk | I haven't seen it's rust until asked if it's possible to build wheels in timely manner :) | 15:39 |
| dansmith | noonedeadpunk: sure, sure :P | 15:39 |
| clarkb | I dont' really care so much about rust. The main thing is I don't have to worry about building it myself which uwsgi hasn't been able to get right for years | 15:40 |
| dansmith | but I mean, gunicorn in a distro package satisfies that requirement right? | 15:40 |
| clarkb | I just wanted to mention it since i know there was discussion about moving away from uwsgi which I think is important given uwsgi's struggles with basic tasks like compiling | 15:40 |
| dansmith | uwsgi is effectively unmaintained anyway, AFAIK | 15:40 |
| clarkb | yes bugfixes only | 15:41 |
| clarkb | re gunicorn sure, but we install from pypi not distro packages typcially (though I could make both work) | 15:41 |
| noonedeadpunk | but point of having same tool/approach for wsgi/asgi is quite relevant today | 15:41 |
| clarkb | gunicron doesn't asgi | 15:41 |
| clarkb | ya exactly | 15:41 |
| dansmith | I guess I just don't want to choose a thing because it's easy and instead pick something that is likely representative of the way people will use it | 15:41 |
| clarkb | I was trying to balance a couple of things since I'm switching anyway | 15:41 |
| dansmith | there's a thing for gunicorn to do asgi AFAIK | 15:42 |
| frickler | looks like a single maintainer project with no distro support? how many years until it gets abandoned, too? | 15:42 |
| noonedeadpunk | clarkb: you can do uvicork for asgi? | 15:42 |
| dansmith | for example, the behavior of mod_wsgi (in rhel) and uwsgi when the application throws an exception on startup is different and that has bitten us, | 15:42 |
| noonedeadpunk | *uvicorn | 15:42 |
| dansmith | where upstream we see and test that behavior but see different stuff at runtime in reality | 15:42 |
| dansmith | frickler: yeah that's my question/concern.. not that it's not awesome now, but will it be awesome in a couple years :) | 15:43 |
| clarkb | noonedeadpunk: right thats not gunicorn though (different package and code base aiui) | 15:43 |
| clarkb | which is totally fine. I'm not advocating for anything here | 15:43 |
| noonedeadpunk | but should not these tools be more applicatiion agnostic? | 15:43 |
| clarkb | I'm more just saying I've got a reason to experiment and I'm doing so. And its really easy to do and I don't think it would be any harder in say devstack | 15:44 |
| noonedeadpunk | clarkb: I'm inteerested to hear out updates :) | 15:44 |
| noonedeadpunk | As also was thinking about renaming/repurposing https://opendev.org/openstack/ansible-role-uwsgi to support different tools while providing same input fromat | 15:45 |
| clarkb | in general I think there is value in tools that do both asgi and wsgi to minimize the amount of double accounting you have to do as things transition (and probably we won't get complete transition in the end in the real world anyawy). And it seems like few tools really try to do both well. But if you know of others I'd be happy to hear about them | 15:48 |
| noonedeadpunk | ++ | 15:48 |
| clarkb | and I agree uwsgi being basically one maintainer didn't turn out well in the end so that is a risk. but it also didn't prevent openstack from unilaterally choosing it | 15:48 |
| clarkb | devstack must be using distro packages for uwsgi? I wonder why it doesn't frequently fail on building it from source if that isn't the case | 15:51 |
| noonedeadpunk | clarkb: well, we run uwsgi and build wheels for it in CI and failures are super rare tbh | 15:52 |
| noonedeadpunk | but we also not taking master branch, only tags | 15:52 |
| clarkb | we build the latest version on pypi from sdist to wheel. Though looking closer the problems we haev are building for aarch64. Maybe that is the difference | 15:53 |
| noonedeadpunk | oh, yes, we don' | 15:53 |
| noonedeadpunk | don't do aarch in CI so far ( | 15:53 |
| noonedeadpunk | so can be difference indeed | 15:54 |
| frickler | clarkb: using distro pkgs for most distros, yes https://opendev.org/openstack/devstack/src/branch/master/lib/apache#L81-L97 | 15:58 |
| mnasiadka | Well, uWSGI supports passing command line arguments to wsgi scripts (where e.g. gunicorn passes it's own CLI arguments to WSGI scripts - and it doesn't work with Nova IIRC, haven't checked other services) - I think due to devstack using uWSGI in it's testing - all deployment projects are sort of either doing or thinking to do the same (so at least they are not on their own if they see issues) | 16:12 |
| mnasiadka | Kolla also uses distro packages for uwsgi (that we're now switching from apache+mod_wsgi to uwsgi - slowly but it's happening) | 16:14 |
| noonedeadpunk | mnasiadka: what is your plan for skyline then? | 16:15 |
| mnasiadka | We run skyline in gunicorn today, probably not going to touch it :) | 16:15 |
| mnasiadka | (at least not in the current iteration) | 16:16 |
| noonedeadpunk | uvicorn, but yeah... | 16:16 |
| mnasiadka | right | 16:16 |
| noonedeadpunk | mnasiadka: btw, have you already moved neutron to uwsgi? as it's been historically very problematic for us | 16:18 |
| noonedeadpunk | and had to revert back to eventlet mul;tiple times | 16:18 |
| mnasiadka | Not yet, and probably we'll wait for it until F to experiment at least half a cycle with that | 16:19 |
| noonedeadpunk | I can recall them being super active with dropping eventlet... | 16:20 |
| noonedeadpunk | so was not sure if there's possibility tbh | 16:21 |
| mnasiadka | it works for now (not counting the fix that is being merged right now with neutron-dhcp-agent and ipv6) | 16:32 |
| clarkb | mnasiadka: one work around for that is to rely on config files or env vars rather than command line args. But that depends on the software supporting it in the first place. That might be feedback for the eventlet removal work since that moves things to asyncio and asgi theoretically? | 16:32 |
| clarkb | (and uwsgi doesn't do asgi) | 16:33 |
| mnasiadka | clarkb: I'm not saying it was impossible to make it work, but I decided to stop going against the stream - just switch to something that is well tested. So I'd assume that for long term - eventlet removal (or some other piece of work) should probably also include switching devstack to some other chosen wsgi server, since uwsgi is not really viable... | 16:45 |
| clarkb | I agree | 16:45 |
| mnasiadka | well, wsgi/asgi | 16:46 |
| clarkb | though as frickler points out the uwsgi install usually comes from the distro so maybe the majority of the problems with not having maintainers are avoided | 16:46 |
| spotz[m] | o/ | 17:05 |
| gouthamr | tc-members: o/ a gentle reminder that we'll be catching up here in ~41 mins for our weekly meeting | 17:19 |
| gouthamr | #startmeeting tc | 18:00 |
| opendevmeet | Meeting started Tue Mar 18 18:00:17 2025 UTC and is due to finish in 60 minutes. The chair is gouthamr. Information about MeetBot at http://wiki.debian.org/MeetBot. | 18:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 18:00 |
| opendevmeet | The meeting name has been set to 'tc' | 18:00 |
| gouthamr | Welcome to the weekly meeting of the OpenStack Technical Committee. A reminder that this meeting is held under the OpenInfra Code of Conduct available at https://openinfra.dev/legal/code-of-conduct. | 18:00 |
| gouthamr | Today's meeting agenda can be found at https://wiki.openstack.org/wiki/Meetings/TechnicalCommittee | 18:00 |
| gouthamr | #topic Roll Call | 18:00 |
| cardoe | o/ | 18:00 |
| noonedeadpunk | o/ | 18:00 |
| mnasiadka | o/ | 18:00 |
| gmann | o/ | 18:00 |
| gtema | o/ | 18:00 |
| frickler | \o | 18:01 |
| bauzas | \o | 18:01 |
| gouthamr | courtesy ping: spotz[m] | 18:02 |
| gouthamr | thank you all for joining; nearly full house today.. lets get started | 18:04 |
| gouthamr | #topic Last Week's AIs | 18:05 |
| spotz[m] | o/ | 18:05 |
| gouthamr | i think we didn't take any concrete AIs, but we have a couple of hanging threads | 18:06 |
| gouthamr | 1) Operator Engagement on Mailing Lists | 18:06 |
| gouthamr | s/lists/list | 18:06 |
| gouthamr | i think post this discussion, we were recalling fungi's email regarding merging openstack-operators with openstack-discuss and shutting down openstack-operators | 18:07 |
| gouthamr | the tl;dr version was that there was a lot of cross posting between the lists and splintering of conversations | 18:08 |
| gouthamr | and the rest of the thread went off into a different issue iiuc | 18:08 |
| gouthamr | #link https://lists.openstack.org/pipermail/openstack-operators/2018-August/015768.html ([Openstack-operators] [all] Bringing the community together (combine the lists!)) | 18:09 |
| fungi | note that mailing list thread was also a continuation of forum sessions from months prior | 18:09 |
| cardoe | despite being an operator (at least I think I qualify?) I don't follow that list. | 18:10 |
| fungi | where we'd first socialized the idea with the various parties affected | 18:10 |
| fungi | cardoe: it hasn't existed since 2019 | 18:10 |
| fungi | or late 2018 really | 18:10 |
| gouthamr | ack, so i wonder if the operators who had this feedback were not participating then, or if they're recaliberating based on how things have worked so far | 18:11 |
| fungi | or they're unfamiliar with the fact that it used to be the case but don't realize the drawbacks that caused us to not continue doing so | 18:11 |
| fungi | chesterton's fence | 18:11 |
| gouthamr | yeah | 18:12 |
| * gouthamr looks up chesterton's fence | 18:12 | |
| * gouthamr :D wow | 18:12 | |
| mnasiadka | I personally think the argument that operators are afraid to ask questions on openstack-discuss is a bit untrue - I see a lot of people asking questions about openstack-ansible or kolla-ansible based deployments recently (although I think I see more of them on Reddit) | 18:13 |
| bauzas | I wonder how many people know that they can be notified differently with the tags | 18:15 |
| gouthamr | hmm, another thing being "developers" don't care about (some of) their problems, and they're sometimes only looking for feedback from fellow operators | 18:15 |
| bauzas | and everytime I see an email without a tag, bah | 18:15 |
| fungi | i would argue that there are more operator questions on openstack-discuss these days than developer discussion | 18:15 |
| bauzas | at least for nova, the ML is the only medium when we ask operators about some solution we would do | 18:16 |
| bauzas | (for example, a deprecation or a virt driver modification) | 18:16 |
| bauzas | so if operators don't read the ML, huh | 18:17 |
| gouthamr | yeah, concerning - there's an operators discussion at the PTG | 18:18 |
| gouthamr | i can add this topic there, and suggest that the TC is looking to understand their concerns | 18:18 |
| gouthamr | #link https://etherpad.opendev.org/p/apr2025-ptg-os-operators (os-operators PTG Etherpad) | 18:20 |
| spotz[m] | Back in the day we got more interaction from the Twitter account but Chris decided to close it. Not that I’d posted anything in awhile | 18:20 |
| gouthamr | ack | 18:21 |
| gouthamr | lets revisit this after we've understood what the issues are | 18:21 |
| gouthamr | thanks for sharing the context | 18:21 |
| gouthamr | the only other AI that i see is the follow up on the OIF board discussions | 18:22 |
| gouthamr | this happened after our meeting, and am assuming everyone caught up on it | 18:22 |
| gouthamr | #link https://openinfra.org/blog/openinfra-linux-foundation-collaboration (jbryce's blog post) | 18:23 |
| gouthamr | #link https://lists.openinfra.org/archives/list/foundation@lists.openinfra.org/thread/577NQIACCIXR72GVWEKTHWF54BHF7ZRG/ (Recap of the March OIF Board Meeting) | 18:24 |
| gouthamr | anything else to be said/shared wrt this? | 18:24 |
| gouthamr | per that update, i think all community members that are foundation members should receive an invitation to vote to ratify this decision of the board | 18:25 |
| gouthamr | "sometime in April" | 18:25 |
| gouthamr | that's all the hanging threads we had last week, which i'm terming AIs :) | 18:26 |
| gouthamr | was anyone else tracking anything else? | 18:26 |
| gouthamr | #topic TC Vice-chair | 18:28 |
| gouthamr | ^ just an announcement that we've made the change, and ty for +1s: | 18:28 |
| gouthamr | #link https://review.opendev.org/c/openstack/governance/+/944174 (Add 2025.2 TC chair and vice-chair) | 18:28 |
| gouthamr | thanks for volunteering noonedeadpunk! | 18:29 |
| noonedeadpunk | I forgot to vote though :D | 18:29 |
| gouthamr | haha, i merged it before you could -1 | 18:29 |
| noonedeadpunk | but my pleasure:) | 18:29 |
| spotz[m] | Hehe | 18:30 |
| gouthamr | :) | 18:30 |
| fungi | you've been press-ganged into service | 18:30 |
| gouthamr | #topic TC Meeting times poll | 18:30 |
| gouthamr | we've polled for meeting time conformance/convenience at the start of TC terms | 18:31 |
| noonedeadpunk | should we maybe start if someboidy not ok with current time? | 18:31 |
| gouthamr | good point | 18:32 |
| gtema | I am not ok | 18:32 |
| gtema | as always was | 18:32 |
| gouthamr | the US DST shift happened two weeks ago, and the EU one is in 1 week | 18:32 |
| noonedeadpunk | ++ then let's make a poll | 18:32 |
| gouthamr | #link https://framadate.org/os-tc-2025-2 | 18:32 |
| frickler | I'm also not happy with the current time fwiw | 18:33 |
| gouthamr | can you please pick from teh options in the poll, as to when you would like to start this one hour meeting | 18:33 |
| noonedeadpunk | gtema: it's still high chance we end up with a very simmilar timeframe, as EU/US is impossible to match and make everybody happy :( | 18:33 |
| gouthamr | ignore the specific dates, and think of the typical days of the week | 18:33 |
| bauzas | that usual meeting time is for me a bit difficult to attend, unless I have chips and beers | 18:33 |
| gouthamr | and please keep the daylight savings changes in your time zones | 18:34 |
| gouthamr | haha, i'd petition for some budget for that bauzas | 18:34 |
| gtema | noonedeadpunk - strongly disagree. Finding good EU/US is very easy | 18:34 |
| gtema | we have multiple meetings a week in friendly time for everyone | 18:34 |
| bauzas | I'm already packed during the golden hour, ie. 4pm-5pm UTC | 18:34 |
| bauzas | so there is no easy win for me | 18:35 |
| frickler | so add in another option: stop doing synchronous(sp?) meetings completely | 18:35 |
| bauzas | I'd maybe vote for no change due to the other large number of meetings I have on 4-6pm UTC days | 18:35 |
| noonedeadpunk | I'm not particulary happy about time as well, but it's easy to attend and no conflicts at least | 18:36 |
| spotz[m] | Ok day matters date doesn’t? | 18:36 |
| gouthamr | i love async work, buuuuut, i think this meeting acts as a good check point to make progress and connect with everyone | 18:37 |
| frickler | spotz[m]: day of week only, yes | 18:37 |
| spotz[m] | I have this blocked so don’t mind keeping. | 18:38 |
| fungi | challenges settling on a convenient time for everyone was one of the reasons we opted to stop holding weekly meetings when i was on the tc | 18:38 |
| spotz[m] | I have to pour through the calendar to move it | 18:38 |
| gmann | Even TC does not need to hold meeting as per mandatory requirement but no meeting can impact on TC working or slow down its activeness. | 18:39 |
| noonedeadpunk | we're not particulary fast even now tbh | 18:39 |
| fungi | (we held office hours, but full tc member attendance was not expected) | 18:39 |
| noonedeadpunk | but it could become way worse, sure | 18:39 |
| gmann | Main idea behind increasing the frequency of meeting (form monthly meeting to weekly) to speed up the TC work and provide fast decision/progress to community ask/need | 18:39 |
| gmann | noonedeadpunk: we used to be :) | 18:40 |
| gouthamr | gmann: dunno if you mean, we were way worse, or if we were particularly fast.. | 18:40 |
| gmann | I am not how many of us or future TC can be online/available to discuss the things | 18:40 |
| bauzas | I can't just fill the framadate now, I need to compare with all my other lovely meetings | 18:41 |
| gouthamr | sure take your time, bauzas | 18:41 |
| fungi | back then we observed that holding meetings every week encouraged us to wait until the meeting to discuss or review things, and so not having those meetings resulted in speeding some things up because people didn't feel compelled to sit on them waiting for a meeting | 18:41 |
| gouthamr | i'll need an answer to collate by next week | 18:41 |
| gmann | I mean having meeting and weekly is one of the important part where we make fast decision | 18:41 |
| gouthamr | ++ | 18:41 |
| gmann | if we think of stopping meeting we should first come up with the better alternate to continue TC speed/activeness etc | 18:42 |
| mnasiadka | Luckily I don't have millions of meetings in the shared EU/US slot - so I'm happy with sort of anything, but I feel frickler/gtema/others that current slot is 7pm EU time, and after DST switch it's going to be 8pm | 18:42 |
| gtema | pings are working very good fwiw | 18:42 |
| gouthamr | yeah, even advancing by an hour can make a difference | 18:42 |
| gmann | gtema: it does some time but not a collective gathering of discussion among all TC members | 18:43 |
| bauzas | honestly I would even propose a late-EU meeting | 18:43 |
| noonedeadpunk | yeah, actually moving an hour before to keep same time witrh DSST makes sense indeed | 18:43 |
| gmann | it is hard to expect every TC members to be online/available at random time considering we all work from different TZ | 18:43 |
| gouthamr | +1 | 18:43 |
| bauzas | this would be easier for me to juggle like when it was at 9pm UTC | 18:43 |
| spotz[m] | That’s what we do for CentOS board Bauzaun | 18:44 |
| spotz[m] | Bah typo, but yeah late means EU folks dinner with family | 18:45 |
| frickler | how did you get german into your auto-correct? ;) | 18:45 |
| gtema | looooool | 18:45 |
| gouthamr | i'll let you think about this and answer the poll.. lets catch up on the results next week | 18:46 |
| gouthamr | #topic A check on gate health | 18:46 |
| spotz[m] | I switch to German keyboard and get stuck | 18:46 |
| * bauzas googles Bauzaun | 18:46 | |
| gtema | construction site fence | 18:47 |
| fungi | not health-related specifically, but we added a new rackspace flex region in nodepool (dfw3), which has doubled the quota for flex nodes from 32 to 64, and we've heard tell of a third region coming online soon too | 18:47 |
| gouthamr | nice ++ | 18:47 |
| frickler | gate is borked for kolla currently due to a bug in neutron, fix is dealing with gate issues, too | 18:47 |
| frickler | #link https://review.opendev.org/c/openstack/neutron/+/944763 | 18:48 |
| frickler | otherwise most release related patches I watched went mostly fine | 18:49 |
| gmann | other thing to note: doc job is migrated to Noble (running on python 3.12) on Friday and we had 3 project failing (a few of them fixed in advance of migration). out of those 3 projects, kolla-ansible fix is not merged. and rest all merged. | 18:49 |
| gmann | #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/JOMDY26TCW7OX3NXRGOYQCIDXNNJ4E25/ | 18:49 |
| fungi | openstack-helm is presumably broken on master for the moment too, since i just helped them merge the openstack-helm-infra repo into it an hour or two ago, but they're aware and working on it | 18:49 |
| frickler | k-a docs fix is also blocked by the neutron issue | 18:49 |
| gmann | I am also monitoring doc job failure if that is related to Noble migration but not seeing any other as of now | 18:49 |
| gmann | #link https://zuul.openstack.org/builds?job_name=openstack-tox-docs&branch=master&result=FAILURE&skip=0 | 18:49 |
| gmann | frickler: yeah | 18:49 |
| mnasiadka | kolla-ansible fix is not merged, because we're waiting for the Neutron fix (and then daily container build job) - once we're back in business we'll merge the doc fix | 18:50 |
| gmann | yeah, thanks | 18:50 |
| gouthamr | the neutron fix will require an RC2? | 18:50 |
| frickler | I'd think so, yes. though kolla is still consuming neutron for master for now | 18:50 |
| mnasiadka | but we can switch after Neutron merges backport of the fix in 2025.1 | 18:51 |
| gouthamr | ack, good to see the RC1 window pass by with relatively less fire fighting.. thanks for the hard work on the releases, and the CI changes | 18:52 |
| gouthamr | gmann: ty for the update on the noble transition bits too! nice that we didn't need to revert to jammy | 18:53 |
| gouthamr | anything else for $topic? | 18:53 |
| gouthamr | #topic PTG Planning | 18:54 |
| gouthamr | i booked some slots: https://ptg.opendev.org/ptg.html | 18:55 |
| gouthamr | ^ the same amount of time as last time, and the same hours | 18:55 |
| gouthamr | but, i was ranting in #opendev-events that we've some project slots on all days, for all four hours :D | 18:56 |
| gouthamr | so it'll be hard to get everyone participating | 18:56 |
| mnasiadka | I might not be able to join PTG sessions - I'll be in NZ timezone at that time | 18:57 |
| gouthamr | ah, ack, hopefully vacationing! | 18:57 |
| mnasiadka | half-vacation (first work, then vacation) ;-) | 18:57 |
| spotz[m] | Yeah I’m here next week then out for Kubecon and PTO during the PTG | 18:58 |
| gouthamr | TC meeting slots: | 18:59 |
| gouthamr | Monday/7th Apr: 1300-1500 UTC | 18:59 |
| gouthamr | Friday/11th Apr: 1500 to 1700 UTC | 18:59 |
| gouthamr | differences from last time: there was a diversity-wg meeting at 1300 UTC on Monday and we met 1400-1600 UTC on Monday | 18:59 |
| gouthamr | based on the topics we add to the etherpad, i can request some more time: | 18:59 |
| gouthamr | #link https://etherpad.opendev.org/p/apr2025-ptg-os-tc (TC PTG Etherpad) | 19:00 |
| gouthamr | ack spotz[m] | 19:00 |
| gouthamr | alright, we've reached the end of the hour | 19:01 |
| gouthamr | no time for Open Discussion | 19:01 |
| gouthamr | but did anyone want to add anything to the minutes here? | 19:01 |
| gouthamr | thank you all for attending, and for the lively discussion. Please do look for some changes that need review here: | 19:02 |
| gouthamr | #link https://review.opendev.org/q/project:openstack/governance+status:open (Open Governance Changes) | 19:02 |
| gouthamr | #endmeeting | 19:03 |
| opendevmeet | Meeting ended Tue Mar 18 19:03:06 2025 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 19:03 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/tc/2025/tc.2025-03-18-18.00.html | 19:03 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/tc/2025/tc.2025-03-18-18.00.txt | 19:03 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/tc/2025/tc.2025-03-18-18.00.log.html | 19:03 |
| mnasiadka | gouthamr: should I have roll-call voting permissions now, or are we waiting for a new cycle or something? | 19:06 |
| fungi | also a heads-up i sent this while the meeting was in progress: https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/L4OG3TJ5JBVS4IS7KCQKXER736PWEITB/ "[administrivia] Recent change in moderation for new subscribers" | 19:08 |
| gouthamr | mnasiadka: o.O my bad, i'll fix that uo | 19:15 |
| gouthamr | mnasiadka: done, please check and tell me if you don't | 19:16 |
| gouthamr | wow that made me look in my Gmail's spam folder | 19:17 |
| gouthamr | fungi: thanks! | 19:17 |
| gouthamr | the blackout curtains seller probably got some clicks | 19:20 |
| fungi | until i blacked them out | 19:21 |
| gouthamr | :D | 19:22 |
| fungi | i saw a total of 4 spam messages from 3 new subscribers, all blocked and cleaned up since | 19:23 |
| mnasiadka | gouthamr: works, thanks! | 19:32 |
| cardoe | For next week I wanted to float out if we wanted to set a goal around adding type hinting to our code bases. | 20:16 |
| fungi | i think some openstack projects have already added a bit? | 20:18 |
| fungi | 100% coverage would be a massive undertaking, but also seems like something projects can do already if they see value in it, since they obviously are | 20:19 |
| fungi | we started trying to annotate zuul as an experiment 3-4 years ago, but decided that it made the code less readable and we wasted more time trying to appease the type checker and keeping up with the shifting sands of its expectations than we saved by finding net new bugs, so eventually gave up and started ripping all the annotations back out | 20:21 |
| fungi | every project/community is different though | 20:23 |
| frickler | IMO it doesn't make sense to propose this if there aren't volunteers interested in doing the work for all repos. if these volunteers are indeed around, I'd like to ask what it would take to get them to look into eventlet-removal instead | 20:26 |
| fungi | i think type annotations are probably far easier if you're writing them into new python projects from the start. trying to annotate existing projects is a more iffy proposition. even upstream in cpython most of the stdlib does not include annotations, and the core devs maintaining the modules are content to let people who want annotations maintain them elsewhere in a separate | 20:26 |
| fungi | typeshed | 20:26 |
| frickler | that's another good point, ack | 20:27 |
| clarkb | cinder and nova both have mypy targets in tox.ini. cinder runs a cinder-mypy job in ci. nova may run mypy as part of pre-commit I'm not sure | 20:28 |
| clarkb | the last time this came up I asked the proposers to seek out those already doing that for freedback and I'm not sure that happened | 20:29 |
| clarkb | but that seems like a good starting place still. Basically ask people who are doing it whether or not they think it is possible and has value and additioanlly for any input on implementation | 20:29 |
| clarkb | I thought pytype looked promising for the use case of people wanting tool integration beacuse it does inference and doesn't require you to have a ton of annotations. That was the theory anyway but I tried to test it on zuul and found there is a lot that it fails to infer and then you're back to manually annotating things | 20:31 |
| clarkb | however, as I annotated certain things it was able to infer more and more so tools like that may represent a middle ground for older code bases where you get the benefits and don't have to do quite as much wor | 20:32 |
| cardoe | I only bring it up because I’ve been chasing down type related bugs in half a dozen projects and trying to land the fix for the release and/or getting back ports to happen. | 23:25 |
| cardoe | Running mypy with some stricter settings exposed the issues right away. | 23:26 |
| cardoe | I was actually thinking about the eventlet removal with the typing annotations. Cause that’s helping me see where changes need to happen. | 23:27 |
| JayF | cardoe: add type annotations to the ironic ptg page; I think it doesn't hurt to add them as we go | 23:27 |
| JayF | cardoe: and while I know doing stuff like that openstack-wide is ideal; I think it'd be interesting to be a pilot program. And maybe the best intern project of all time (for an outreachy/MLH/college intern project) | 23:28 |
| cardoe | Yeah great point. | 23:28 |
| JayF | I pushed hard for the pydoc w/typing we have now a while back | 23:29 |
| JayF | when I discovered IDEs could consume them | 23:29 |
| JayF | so there are even places where we may just be able to move the documentation into code instead of having to discover something net-new | 23:29 |
| cardoe | Yeah that’s been nice. | 23:29 |
| JayF | with maybe an `| None` added here or there :) | 23:29 |
| cardoe | We can’t have nice things like that. I’m being grumped at for using Python 3.7 syntax by someone. | 23:30 |
| cardoe | 3.10 where that appeared would just be bridge too far. | 23:31 |
| clarkb | I personally dislike adding special syntax that only newer pythons support for things that are doable otherwise. I say this as someone with the oldest python installed being 3.11. Simply because it has happend multiple times where I go to run something and it can't because of a syntax issue and nothing fundamental to what is being accomplished | 23:32 |
| cardoe | Well it’s enterprise Linux folks that have 3.6 somewhere. | 23:32 |
| cardoe | But anyway…. len(None) doesn’t work. /tableflip | 23:33 |
| cardoe | Pydoc does say it’s a string. | 23:34 |
| clarkb | side note python3.13 repl added colors | 23:35 |
| cardoe | The mock always does “” | 23:36 |
| cardoe | Just a thought of something we could encourage. Certainly could encourage it on newer projects. It doesn’t need to be on everything. Only when inference doesn’t work. Much like Rust. | 23:39 |
| clarkb | ya thats why I was hopeful for pytype over mypy | 23:40 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!