| opendevreview | Roja Eswaran proposed openstack/diskimage-builder master: debootstrap: add DIB_MMDEBSTRAP_EXTRA_ARGS support https://review.opendev.org/c/openstack/diskimage-builder/+/999989 | 14:52 |
|---|---|---|
| opendevreview | Julia Kreger proposed openstack/diskimage-builder master: Add dib-init-image tool for creating empty disk images https://review.opendev.org/c/openstack/diskimage-builder/+/1001740 | 19:07 |
| opendevreview | Steve Baker proposed openstack/diskimage-builder master: Add dib-init-image tool for creating empty disk images https://review.opendev.org/c/openstack/diskimage-builder/+/1001740 | 19:09 |
| opendevreview | Roja Eswaran proposed openstack/diskimage-builder master: debootstrap: add DIB_MMDEBSTRAP_EXTRA_ARGS support https://review.opendev.org/c/openstack/diskimage-builder/+/999989 | 20:00 |
| opendevreview | Merged openstack/diskimage-builder master: Add minimum pyproject.toml https://review.opendev.org/c/openstack/diskimage-builder/+/1000692 | 20:07 |
| opendevreview | Merged openstack/diskimage-builder master: Bump upper version of flake8 https://review.opendev.org/c/openstack/diskimage-builder/+/1000680 | 20:25 |
| TheJulia | Looks like DIB is blocked on glean being used: https://review.opendev.org/c/opendev/glean/+/1003156 | 21:11 |
| clarkb | TheJulia: do you have poitners to the underlying issue? setup.py should work just fine with modern python packaging (which I noted on my review of 1003156) | 21:27 |
| clarkb | I don't think adding a pyproject.toml to an PBR project today should fix any issues | 21:28 |
| clarkb | what it does do is give you more control over the package build environment | 21:28 |
| TheJulia | claude was able to reproduce locally because my setuptools is like 80 locally | 21:28 |
| TheJulia | uhh, one moment | 21:28 |
| clarkb | please avoid telling me to ask claude | 21:29 |
| clarkb | I do not have a claude subscription | 21:29 |
| TheJulia | https://zuul.opendev.org/t/openstack/build/c38e1a8404094a0daa8ad59ea75b361f/log/diskimage-rocky-container.log <-- search for tomli | 21:29 |
| TheJulia | I wasn't suggesing that | 21:29 |
| clarkb | I am happy to debug issues if they are shared directly | 21:29 |
| TheJulia | I was noting it was able to reproduce it directly locally | 21:29 |
| clarkb | TheJulia: I believe that error is the error related to the PBR releases that happend recently (the log is from a number of days ago) | 21:30 |
| clarkb | I don't think any of those issues should persist at this point (releases were pulled and stephenfin did extra work to fix things up) | 21:30 |
| TheJulia | so do you believe just rechecking again should clear, becuase I noticed on another change that they were rechecked this morning and also failed | 21:30 |
| clarkb | yes, the problem was that PBR was using pyproject.toml prpcessing code which depends on a toml serialization library regardless of whether or not the project even had a pyproject.toml | 21:31 |
| TheJulia | oh, no, I misread the log, let me recheck then | 21:31 |
| clarkb | I believe this has been corrected since it broke everything | 21:31 |
| clarkb | and well now everything isn't broken anymore. It is possible there are still some edge cases | 21:31 |
| TheJulia | okay, change rechecked, so time will tell | 21:32 |
| clarkb | essentially if you are trying to use pyproject.toml then you need to haev a new enough toolchain to do so. Having a project.toml build spec can help ensure you do by ensuring setuptools etc are new enough. But if you don't have a pyproject.toml at all then none of that should matter and you should fall into the compatibility code on modern systems or the old school tools will | 21:33 |
| clarkb | do their things per usual on older systems | 21:33 |
| clarkb | the bug was that PBR assumed there would be a toml library regardless | 21:33 |
| TheJulia | That would do it | 21:34 |
| TheJulia | Looks like it was just pbr related, at least in CI so good for now. | 22:17 |
| TheJulia | I'll abandon the other change | 22:17 |
| clarkb | one note is that I think the proposed change breaks glean too. The CI jobs on the change seem to catch the problem | 22:24 |
| clarkb | specifically there is an entrypoint that has gone missing I think | 22:24 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!