Opened 7 years ago
Last modified 7 years ago
#29125 closed Bug
BUG: Q object deconstruct is inconsistent when passing multiple kwargs. — at Version 1
| Reported by: | Harro | Owned by: | nobody |
|---|---|---|---|
| Component: | Database layer (models, ORM) | Version: | 2.0 |
| Severity: | Release blocker | Keywords: | |
| Cc: | hvdklauw@… | Triage Stage: | Accepted |
| Has patch: | yes | Needs documentation: | no |
| Needs tests: | no | Patch needs improvement: | no |
| Easy pickings: | no | UI/UX: | no |
Description (last modified by )
Here is a branch with a randomly failing test that proves this: https://github.com/hvdklauw/django/blob/bug/q_destruct/tests/queries/test_q.py#L63
The biggest issue is that now makemigrations is detecting changes since we upgraded to django 2.0 (only with python 3.5) that aren't changes at all, just reordered kwargs on the Q objects in out limit_choices_to on some foreignkeys.
This also means we randomly can't commit/release because we have pre-commit hooks and CI that runs makemigrations --check --dryrun
Upgrading to python 3.6 would fix it (because of ordered kwargs) but as Ubuntu 16.04 is still the latest LTS, python 3.5 is what we are stuck with.
This is the commit that broke it: https://github.com/django/django/commit/508b5debfb16843a8443ebac82c1fb91f15da687