21:06:49 <timburke> #startmeeting swift
21:06:49 <opendevmeet> Meeting started Wed Dec  7 21:06:49 2022 UTC and is due to finish in 60 minutes.  The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:06:49 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:06:49 <opendevmeet> The meeting name has been set to 'swift'
21:07:05 <timburke> looks like we've already got people here for a swift meeting :-)
21:07:55 <mattoliver> yup :)
21:08:10 <timburke> i neglected to update the agenda, but mostly i just wanted to give updates on last week's items, anyway
21:08:29 <timburke> #topic leading slashes bug
21:09:46 <timburke> clayg pushed back on continuing to hack do pre/post processing, instead thinking that just inlining (and fixing) parse_request may be better
21:10:08 <timburke> i finally got around to trying that approach -- and i'm torn about which one i prefer
21:10:35 <timburke> #link https://review.opendev.org/c/openstack/swift/+/866801
21:10:53 <timburke> ^^^ inlines and fixes up parse_request
21:11:02 <timburke> #link https://review.opendev.org/c/openstack/swift/+/863441
21:11:26 <timburke> ^^^ undoes the change with some post-processing
21:11:42 <mattoliver> I guess there is a bigger maintainence burden on the latter approach.. but the other is trying to fix things in potentually odd spots.
21:12:22 <mattoliver> either way, the initial refactor http protocol stuff is nice :)
21:12:39 <timburke> the inlining did give me the opportunity to clean up some other hacks we had (specifically, by addressing https://bugs.python.org/issue33973 in a much more direct way)
21:13:35 <timburke> and the function itself hasn't seen *that many* updates in the past ten years or so, looking at main
21:14:34 <mattoliver> OK well that's promising. Not too much of a maintence burden if it doesn't change too much
21:14:59 <acoles> timburke: thanks for doing more work! I haven't had chance to look at the second approach yet
21:15:29 <timburke> otoh, there's a whole bunch more branches now present in our code that are likely untested, and it seems unlikely that we'd notice any upstream changes to the function in the future (so when there's inevitably another patch in a couple years, we won't even know to think about whether to apply it)
21:17:03 <mattoliver> need a monthly zuul git diff job :P
21:17:47 <acoles> mattoliver: volunteering ;-)
21:17:59 <timburke> mattoliver, there's enough of a delta between us and upstream that i'm not sure how helpful that would be... though maybe diffing between upstream revisions could work well enough?
21:18:25 <mattoliver> well maybe no diff from what we have, but from current master of eventlet is.
21:18:33 <mattoliver> or whatever. so we can detect changes
21:18:49 <mattoliver> or a git blame, or just an md5 check
21:18:56 <mattoliver> (me is thinking out loud)
21:19:19 <timburke> i also proposed a couple follow-ons -- the first moves another stdlib-workaround to parse_request, so we don't have to repeat a bunch of work that eventlet's doing; the second removes the need for the workaround entirely
21:21:41 <timburke> so if you get a chance, take a look at the new approach, see which way you prefer. i think we *do* still need to do *something* -- there are just too many versions of python in the wild that break us, and no traction from upstream to fix it
21:22:25 <timburke> and it'd be nice to have even our non-voting jobs green as we do some other CI maintenance stuff
21:22:29 <timburke> speaking of...
21:22:33 <timburke> #topic pytest
21:23:20 <timburke> still no further feedback on the requirements change that's blocking us moving to pytest, which in turn is blocking us from testing on py310 in the gate
21:23:34 <timburke> #link https://review.opendev.org/c/openstack/requirements/+/863581/
21:24:00 <timburke> i'll continue to badger people about that
21:24:19 <mattoliver> thanks for that
21:24:34 <timburke> but i also recently refreshed some gate results, and it looks like i might need to respin a bit anyway (some dependency errors around py2 probe tests)
21:25:06 <mattoliver> ahh py2 :(
21:25:17 <kota> :/
21:25:42 <timburke> #topic py311 segfault
21:26:36 <timburke> upstream *did* respond quickly on https://github.com/python/cpython/issues/99886 -- the fix is merged, so it looks like there's just the one upstream release affected
21:26:49 <mattoliver> oh cool
21:27:04 <kota> nice
21:27:23 <mattoliver> great job timburke
21:27:25 <timburke> unit tests still fail on py311, though, so some of https://review.opendev.org/c/openstack/swift/+/866051 will still be necessary (just not necessarily the __slots__)
21:28:30 <timburke> i'll respin it soon; it should be even more targeted now that https://review.opendev.org/c/openstack/swift/+/866239 has landed
21:29:17 <timburke> that's all i've got
21:29:21 <timburke> #topic open discussion
21:29:29 <timburke> anything else we should bring up this week?
21:31:57 <timburke> in that case, we'll wrap up early :-)
21:32:09 <timburke> thank you all for coming, and thank you for working on swift!
21:32:14 <timburke> #endmeeting