| *** amoralej|off is now known as amoralej | 06:23 | |
| *** tkajinam is now known as Guest5435 | 07:14 | |
| *** Guest5435 is now known as tkajinam | 07:36 | |
| elodilles | fungi: thanks for the info! i saw that you reenqueued the job 2 times and the second time it succeeded with the openstack-upload-github-mirror + release-openstack-python, but failed with announce-release and propose-update-constraints | 08:51 |
|---|---|---|
| elodilles | as i see those have the same problem: "The task includes an option with an undefined variable. The error was: 'dict object' has no attribute 'tag'" | 08:54 |
| elodilles | so I guess we should not release until that issue is not fixed | 08:54 |
| elodilles | hberaud ttx : fyi ^^^ | 08:54 |
| hberaud | ack | 08:55 |
| ttx | ack | 09:44 |
| *** dviroel|out is now known as dviroel | 11:27 | |
| fungi | elodilles: that was a result of how i reenqueued it, second time was still not quite right) and not the jobs themselves | 11:28 |
| fungi | the tool we use for reenqueuing refs seems to have changed its behavior and expected options | 11:29 |
| fungi | normally triggered releases should work fine | 11:29 |
| *** amoralej is now known as amoralej|lunch | 12:08 | |
| elodilles | fungi: ack, thanks! | 12:20 |
| fungi | the way i reenqueued the tag, for some reason it populated the branch field with the tag number instead of providing a tag field, which seems to have worked fine for the sdist/wheel build and upload, but not for the release announcement and constraints update | 12:23 |
| elodilles | i see | 12:25 |
| elodilles | hberaud ttx : ^^^ we can continue to review/approve release patches | 12:27 |
| opendevreview | Stephen Finucane proposed openstack/reno master: Don't attempt to pull from empty cache file https://review.opendev.org/c/openstack/reno/+/850151 | 12:50 |
| hberaud | ack | 12:53 |
| *** amoralej|lunch is now known as amoralej | 13:00 | |
| *** haleyb_ is now known as haleyb | 14:02 | |
| opendevreview | Boxiang Zhu proposed openstack/releases master: Zed-2 Release for skyline-apiserver and skyline-console https://review.opendev.org/c/openstack/releases/+/850361 | 15:16 |
| opendevreview | Boxiang Zhu proposed openstack/releases master: Zed-2 Release for skyline-apiserver and skyline-console https://review.opendev.org/c/openstack/releases/+/850361 | 15:28 |
| *** akekane_ is now known as abhishekk | 15:37 | |
| *** amoralej is now known as amoralej|off | 15:46 | |
| ttx | The skyline beta release is failing check due to PyPI teasts failing, so I suspect it's because skyline-apiserver and skyline-console are not set up on PyPI | 15:51 |
| ttx | Not sure how that's usually handled though... I thought that was auto-created honestly :) fungi might remember | 15:52 |
| ttx | "could not find users with permission to upload packages for skyline-apiserver. Is the sdist name correct?" | 15:53 |
| ttx | "README check for openstack/skyline-apiserver failed" | 15:53 |
| ttx | "could not find users with permission to upload packages for skyline-console. Is the sdist name correct?" | 15:53 |
| ttx | "README check for openstack/skyline-console failed" | 15:53 |
| *** marios is now known as marios|out | 15:58 | |
| fungi | pypi (warehouse) creates the project the first time a distribution is uploaded | 16:01 |
| fungi | normally we encourage projects to get their python packaging in shape up front and request an initial alpha-type prerelease so that happens and the name doesn't get taken out from under them while they're getting into shape for a proper release | 16:02 |
| fungi | you can manually make an empty dist with just enough metadata to set a version number like 0 or 0.0.0 and then upload that with twine, but that's discouraged because it's basically getting around the desire for people not to squat names indefinitely on pypi | 16:04 |
| fungi | i wonder if those checks have been added recently and baked in an expectation that projects already exist? | 16:05 |
| elodilles | ttx fungi : as i understand skyline team needs to register their deliverables first at pypi as described here: https://docs.opendev.org/opendev/infra-manual/latest/creators.html#give-opendev-permission-to-publish-releases | 18:25 |
| elodilles | after that is done i guess the validator should pass | 18:25 |
| fungi | elodilles: i suppose that will work. seems like my suggestion to stop squatting names on pypi and just let twine create them on first release never got taken up | 18:28 |
| fungi | and we've unfortunately baked an assumption into openstack's release validation scripts that people will namesquat on pypi | 18:29 |
| fungi | we should probably move that section out of the infra manual and into some openstack-specific doc somewhere, since it's strictly not necessary to do that unless you're trying to satisfy openstack's release validation | 18:29 |
| clarkb | fungi: was the squatting done to ensure the right accounts had privileges to write to the package? | 18:30 |
| clarkb | put another way some packages don't have openstackci as the full admin on the package iirc | 18:30 |
| clarkb | maybe this was a way to achive that | 18:30 |
| fungi | originally pypi required you to create the project manually before you could upload. then they added a create-on-upload workflow by default, and after that they took away the ability to manually create new projects | 18:30 |
| fungi | now the only way to create a new project on pypi is to upload a distribution, in order to make namesquatting there harder | 18:31 |
| clarkb | right, but I think it assumes the account pushing should be admin. If CI pushes it would always be admin which I'm not sure is common for openstack pacakges. Just wondering if maybe that was the reason behind not having it all happen autoamtically | 18:33 |
| elodilles | to tell you the truth i don't know the process, but what i see is our validator fails when checking the permissions, so yes, our validator expects that the package needs to be registered 1st on pypi: https://zuul.opendev.org/t/openstack/build/531ecdc448f04089b25a6917b6bca493/log/tox/validate-request-results.log#124-160 | 18:34 |
| ttx | what's weird is that I found a skyline-console on pypi search but then clicking on it goes nowhere | 18:56 |
| elodilles | oh, so it's either namesquatted already (according to fungi's explanation) or the skyline team started the process some time | 18:58 |
| elodilles | hmm, skyline-console 0.1.0 was added ~1 year ago | 18:59 |
| elodilles | and yes, unfortunately it gives 404 when clicking on it | 19:00 |
| elodilles | so we don't even know any detail about the package | 19:00 |
| elodilles | (on the other hand skyline-apiserver does not exist at least...) | 19:01 |
| *** Guest5201 is now known as diablo_rojo_phone | 20:57 | |
| *** dviroel is now known as dviroel|dinner | 22:22 | |
| *** dviroel|dinner is now known as dviroel|out | 23:20 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!