| opendevreview | Rodolfo Alonso proposed openstack/requirements master: Bump jsonschema to 4.6.0 https://review.opendev.org/c/openstack/requirements/+/845557 | 08:47 |
|---|---|---|
| *** tosky_ is now known as tosky | 11:11 | |
| opendevreview | Rodolfo Alonso proposed openstack/requirements master: Bump warlock to 2.0.1 https://review.opendev.org/c/openstack/requirements/+/849102 | 11:42 |
| opendevreview | Hervé Beraud proposed openstack/requirements master: bump dogpile.cache to 1.1.8 https://review.opendev.org/c/openstack/requirements/+/849138 | 18:25 |
| fungi | prometheanfire: what's the easiest way to figure out why lockfile is still in the upper-constraints.txt? the propose-updates job doesn't seem to log pip's dependency resolution decisions these days | 19:17 |
| prometheanfire | fungi: I don't know an easy way | 19:18 |
| fungi | :/ | 19:18 |
| fungi | i wonder how quickly running generate-constraints would kill my workstation | 19:18 |
| prometheanfire | it's not too bad really | 19:18 |
| fungi | lockfile has been abandoned for a very long time, and we assumed maintainership of it on pypi when we needed to fix a bug in 2015, but also removed it from master global-requirements as soon as possible after that | 19:19 |
| fungi | unfortunately, it seems something we're depending on still declares it as a requirment | 19:19 |
| fungi | so i'm hoping to work out what that something might be | 19:20 |
| prometheanfire | | openstack/tripleo-ci-health-queries | requirements.txt | 109 | lockfile==0.12.2 | | 19:24 |
| prometheanfire | maybe | 19:24 |
| fungi | but tripleo-ci-health-queries isn't in global-requirements.txt either | 19:28 |
| prometheanfire | ya, just doing a beagle search is all | 19:30 |
| prometheanfire | swift-24.0.0/lib/python3.7/site-packages/pip/_vendor/cachecontrol/caches/file_cache.py: from lockfile import LockFile | 19:31 |
| fungi | oh fun | 19:31 |
| prometheanfire | fungi: looks like ^ may be it (did a grep for lockfile piped into a grep for import | 19:31 |
| prometheanfire | recursive on the OSA venv dir | 19:32 |
| fungi | i wonder what version of pip that is | 19:32 |
| fungi | can you do a pip list in that env? | 19:32 |
| fungi | should report the pip version | 19:33 |
| fungi | or i guess just pip --version in there | 19:33 |
| prometheanfire | 21.3.1 | 19:33 |
| fungi | huh... | 19:33 |
| fungi | though also, that's a vendored copy of cachecontrol so in theory it's got an accompanying lockfile implementation which isn't something we're installing separately, so wouldn't get downloaded or listed in the constraints | 19:34 |
| fungi | so i don't think that's where it's coming from either | 19:34 |
| prometheanfire | https://github.com/pypa/pip/blob/main/src/pip/_vendor/cachecontrol/caches/file_cache.py#L77 | 19:35 |
| prometheanfire | ya, not sure | 19:35 |
| prometheanfire | hmm, lockfile isn't in https://github.com/pypa/pip/tree/main/src/pip/_vendor | 19:36 |
| fungi | yeah, i think it's just a module in cachecontrol, not the lockfile distribution | 19:36 |
| prometheanfire | ah | 19:36 |
| fungi | my <something> is too new for generate-constraints, probably a too-new setuptools: | 19:37 |
| fungi | error in anyjson setup command: use_2to3 is invalid. | 19:37 |
| fungi | i'll try doing it with python3.9 instead of 3.10 and see if that's any better | 19:38 |
| fungi | no luck, building my own python interpreters and venvs with latest everything has its drawbacks. i'll need to set up a chroot or something to run this in | 19:50 |
| fungi | i don't seem to be able to pin setuptools back to <58 in order to work around it | 19:50 |
| fungi | argh, i think tox -e generate is just outright broken because anyjson doesn't work with setuptools>=58 and i can't seem to work out how to force it to not use newer setuptools, even on an ubuntu focal machine | 20:24 |
| fungi | okay, i gave up and just removed anyjson from global reqs as a workaround for now | 20:36 |
| fungi | systemd-python also fails to generate its metadata | 20:41 |
| fungi | and suds-jurko | 20:43 |
| fungi | psycopg2 apparently lacks a suitable binary wheel and needs to compile against stuff i don't have installed | 20:54 |
| fungi | as does python-nss but i can't tell what's dragging that in (it's not in global reqs) | 21:09 |
| opendevreview | Jeremy Stanley proposed openstack/requirements master: DNM: Pin nonexistent lockfile version https://review.opendev.org/c/openstack/requirements/+/849149 | 21:20 |
| fungi | new approach ^ | 21:20 |
| opendevreview | Merged openstack/requirements master: bump dogpile.cache to 1.1.8 https://review.opendev.org/c/openstack/requirements/+/849138 | 22:10 |
| opendevreview | Jeremy Stanley proposed openstack/requirements master: DNM: Pin nonexistent lockfile version https://review.opendev.org/c/openstack/requirements/+/849149 | 23:09 |
| fungi | looks like none of the check jobs actually try installing global-requirements.txt, instead they install upper-constraints.txt, so that didn't help. i'm trying to make them run generate-constraints now in the dnm change to see if that gets me anything useful, though i have a feeling it will hit the same problems i've been running into trying to run generate-constraints locally | 23:14 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!