Friday, 2026-02-27

opendevreviewMerged openstack/oslo.utils master: Refactor force_reraise  https://review.opendev.org/c/openstack/oslo.utils/+/97769200:08
opendevreviewMerged openstack/oslo.db master: Fix unit test failure with SQLAlchemy 2.1  https://review.opendev.org/c/openstack/oslo.db/+/97754300:11
opendevreviewMerged openstack/oslo.cache master: Replace remaining usage of deprecated options  https://review.opendev.org/c/openstack/oslo.cache/+/97749200:25
opendevreviewMerged openstack/oslo.cache stable/2025.2: Fix type mismatch  https://review.opendev.org/c/openstack/oslo.cache/+/97504500:25
opendevreviewMerged openstack/oslo.db master: Replace usage of undocumented logging.WARN  https://review.opendev.org/c/openstack/oslo.db/+/97753200:37
opendevreviewMerged openstack/oslo.context master: Parse domain_name header  https://review.opendev.org/c/openstack/oslo.context/+/97537700:44
opendevreviewMerged openstack/oslo.policy master: Replace usage of undocumented logging.WARN  https://review.opendev.org/c/openstack/oslo.policy/+/97753300:52
opendevreviewMerged openstack/taskflow master: Add Gelera specific error code for connection retries  https://review.opendev.org/c/openstack/taskflow/+/97754900:55
opendevreviewMerged openstack/oslo.config master: Replace usage of undocumented logging.WARN  https://review.opendev.org/c/openstack/oslo.config/+/97753100:57
opendevreviewMerged openstack/oslo.messaging master: Replace deprecated datetime.datetime.utcnow  https://review.opendev.org/c/openstack/oslo.messaging/+/97742201:01
opendevreviewMerged openstack/taskflow master: Create separate dependency group per database backend  https://review.opendev.org/c/openstack/taskflow/+/97755001:03
opendevreviewMerged openstack/osprofiler master: Replace deprecated datetime.datetime.utcnow  https://review.opendev.org/c/openstack/osprofiler/+/97742301:08
tkajinamos_api_ref/tests/base.py:129:9: error: Value of type variable "UseFixtureT" of "useFixture" of "TestCase" cannot be "Timeout"  [type-var]13:57
tkajinamos_api_ref/tests/base.py:130:9: error: Value of type variable "UseFixtureT" of "useFixture" of "TestCase" cannot be "OutputStreamCapture"  [type-var]13:57
tkajinamhmm13:57
dansmithstephenfin: on the ContainerFileInspector patch, are you opposed to that being a patch on its own before anything uses it in the later patches (re: your "nothing is setting _inner_format" comment)?15:31
stephenfindansmith: nope, I just missed the _inner_format usage in the next patch (I did look)20:15
dansmithstephenfin: okay cool, and .. is the *args complaint because it messes up the typing or something? I think it's a fairly common pattern for hooking20:16
dansmithI can make it explicit (and then change it again in the next patch), I'm just wondering if that's preference or some actual reason20:18
stephenfinyeah, type checking. if you change it to `*args, **kwargs` you lose the ability to have mypy/ty/pyre/... catch invalid stuff20:20
dansmithokay20:21
stephenfinbefore you ask, I'm fine with `data` in the next patch too. I was hoping it was a limited list but if that's not going to be the case, we're going to need something fairly free form20:21
dansmithack :)20:21
stephenfinwe could do a mapping of formats to parameters for same (rather than a flat list and format prefixes) but I guess that won't work if the same param works for multiple formats20:22
stephenfin(as in dict[str, dict[str, str]])20:22
stephenfintkajinam: looks like my patch to bump testtools merged....and some of their hints are wrong20:23
dansmithyeah, I'd really like it if I didn't have to do the data thing at all, but we need to decrypt the key as soon as we have it so we can know what block of data to grab (and potentially later blocks for the inner format) we kinda need to have it upfront20:24
dansmiths/we have it/we have the LUKS header20:24
dansmithstephenfin: while I have you, completely unrelated...20:25
dansmithI've been manually hacking tox.ini to run coverage on functional tests .. is there some reason we can't/don't just change that globally to run on all the tests?20:26
dansmith(I assume you're the (or a) primary opinion-holder here)20:26
stephenfinno reason: melwitt actually has a patch up to do that for nova you might want to hit https://review.opendev.org/c/openstack/nova/+/97534620:27
dansmithaha, nice.. I have been heckling her for a few missed coverage things by doing said local hack, so I'd have assumed that was why she was proposing it, if not for the other justification20:28
melwitthaha yeah I found the same thing lately a lot of code coverage is coming from func tests only so I was thinking we should include them in coverage. I am concerned that I didn't pick a high enough job timeout on that though20:28
dansmithmelwitt: yeah, a couple times recently I've pecked out a gerrit comment of "you're not testing this" and then realized "well, maybe she is" and ran locally20:29
melwittit actually came out biggest for me recently on a unified limits fix I was working on, showed really low code coverage bc almost all unified limits tests are func tests20:29
dansmithmelwitt: so that runs unit and then functional and then generates the report?20:30
melwittyeah. it's pretty slow though and I wasn't sure if someone like stephenfin might know some trick to make it faster20:30
dansmithI shall do the same in glance20:31
melwittthe .stestr.conf has a default to run unit tests, that's how that has been done20:31
dansmithright20:32
melwittso I just left that alone and added a line to run func tests to the testenv:cover itself20:32
dansmithyeah I didn't see a good way to do that otherwise, which is why I didn't propose a change, but I guess I didn't think about two separate runs adding to the same report20:33
stephenfinmelwitt: Not yet. I know the test I have to verify the schema coverage is slow (sean-k-mooney has complained about it more than once), but I think tacking on coverage just makes _everything_ slower20:33
stephenfinthat's the one I've been seeing timeout but I _think_ it's just an artifact of being the longest-running test and most likely to get caught by the closing door20:34
stephenfin*unit test20:34
melwittyeah, that's kinda what I found in googling is just, coverage adds a lot of overhead period20:34
stephenfindansmith: you should also be able to just pass `--test-path ./glance/tests` to stestr20:34
sean-k-mooneystephenfin: i have forgoten if i have20:34
dansmithstephenfin: yeah that's the local hack I was doing20:35
sean-k-mooneymaybe i rased it once or twice a while ago 20:35
stephenfinonce there's an __init__.py in the tests dir, stestr will do discovery just fine20:35
stephenfinack20:35
melwittI'm gonna apply your suggested edit stephenfin 20:38
stephenfinmelwitt: maybe do it in a follow-up? idk if that exclude regex needs updating if we also include unit tests20:39
dansmithyeah, seems better in the nova case to re-use the commands from the other env20:40
melwittoh. oops20:40
dansmithglance doesn't have any to skip that I know of, so it's simpler20:40
stephenfindoing it in one run would definitely be faster since stestr doesn't need to collect tests a second time (and python startup time is slow)20:41
stephenfinbut like I said, it's probably minimal compared to general run time20:41
stephenfin*as a proportion of total run time20:41
melwittit might be better to include the functional db tests though. api and notification samples are safer to exclude I think i.e. probably don't bring much coverage value20:44
stephenfinmelwitt: definitely one to debate in a follow-up20:47
stephenfin:)20:47
stephenfintkajinam: https://github.com/testing-cabal/testtools/pull/57420:49
melwittstephenfin: hm, it looks like there might be something bad about running unit and func concurrently, lots of weird errors https://review.opendev.org/c/openstack/nova/+/975346 so I'm gonna revert back to PS1 23:34
melwitt*PS223:35

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!