31372 INFO [######+++++++++++...] > Install project via pip...
43120 WARNING ERROR: In --require-hashes mode, all requirements must have their versions pinned with ==.
Sounds related to something that changed in pipās behavior maybe ?
EDIT: Okay i run a test by myself and found it, too:
425142 DEBUG Collecting tablib[html,ods,xls,xlsx,yaml]>=3.0.0
425144 WARNING ERROR: In --require-hashes mode, all requirements must have their versions pinned with ==. These do not:
425146 WARNING tablib[html,ods,xls,xlsx,yaml]>=3.0.0 from https://www.piwheels.org/simple/tablib/tablib-3.1.0-py3-none-any.whl#sha256=41f3950bb717a7beb857ffe1ed8c18a2ea7dbb8549400446c48ff3533327d407 (from django-import-export==2.6.1->-r /opt/yunohost/pyinventory/requirements.txt (line 48))
426266 DEBUG + ynh_exit_properly
I had a look at the last run : went to the repository, clicked the āIntegration level 0ā badge at the top of the README, clicked on the ā2 days agoā right to the Stable line, ended up on YunoRunner for CI then I checked the logs for job #4169 (well thatās indeed a bit not obvious if you donāt know about the first two links ā¦ but anyway, you can also go to YunoRunner for CI and click your app and check the last failed log
Itās not that it takes days to schedule, itās more like the test is pending while other tests are being ran
You can see the queue on the ci home page https://ci-apps.yunohost.org/ (main CI) or https://ci-apps-dev.yunohost.org/ (dev CI) and these days thereās indeed a shitload of pending jobs and the CI canāt keep up (+ some technical issue making jobs crash ugh, gotta find some volunteer with time to dig into it)
titus offered to trigger jobs on his own CI in the meantime
Indeed, Iām very new to the Ynh packaging world, but was impressed by how cluttered the CI queue was I try not forgetting what Iām testing while waiting for the result ^^
Thatās very nice !
Arenāt Github actions free for open source projects ?
Sure (apart that it increases our dependency to Github ) but that wouldnt solve the issue I believe : ultimately testing an app is a lengthy process and itās not unusual that the full chain takes at least 1 hour (for many technical reasons) - and I believe that github actions are limited to 1 concurrent worker accross the entire organization. Hence you would get the same issue of too-many-job-pendings accross the 200~300 app repositories all in yunohost-apps org
Thatās true, but we could then run the CI in our own repo and submit it to the official CI (GitHub or not) only once we are sure our app reaches the target level