| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
Ever since the beginning of Django's migration framework, there's been a
bit of an inconsistency on how index_together and unique_together values
have been stored on the ModelState[^1].
It's only really obvious, when looking at the current code for
`from_model()`[^2] and the `rename_field()` state alteration code[^3].
The problem in the autodetector's detection of the `*_together` options
as raised in the ticket, reinforces the inconsistency[^4]: the old value
is being normalized to a set of tuples, whereas the new value is taken
as-is.
Why this hasn't been caught before, is likely to the fact, that we
never really look at a `to_state` that comes from migration operations
in the autodetector. Instead, in both usages in Django[^5], [^6] the
`to_state` is a `ProjectState.from_apps()`. And that state is
consistently using sets of tuples and not lists of lists.
[^1]: https://github.com/django/django/commit/67dcea711e92025d0e8676b869b7ef15dbc6db73#diff-5dd147e9e978e645313dd99eab3a7bab1f1cb0a53e256843adb68aeed71e61dcR85-R87
[^2]: https://github.com/django/django/blob/b1ffa9a9d78b0c2c5ad6ed5a1d84e380d5cfd010/django/db/migrations/state.py#L842
[^3]: https://github.com/django/django/blob/b1ffa9a9d78b0c2c5ad6ed5a1d84e380d5cfd010/django/db/migrations/state.py#L340-L345
[^4]: https://github.com/django/django/blob/b1ffa9a9d78b0c2c5ad6ed5a1d84e380d5cfd010/django/db/migrations/autodetector.py#L1757-L1771
[^5]: https://github.com/django/django/blob/2351c1b12cc9cf82d642f769c774bc3ea0cc4006/django/core/management/commands/makemigrations.py#L215-L219
[^6]: https://github.com/django/django/blob/2351c1b12cc9cf82d642f769c774bc3ea0cc4006/django/core/management/commands/migrate.py#L329-L332
|
|
Previously, `_generate_plan()` relied on list membership checks,
resulting in quadratic behavior as the plan grew. On large migration
graphs this became a significant performance bottleneck.
This change uses `OrderedSet` for the plan, reducing the complexity to
linear while preserving insertion order and behavior.
Co-authored-by: Nick Pope <nick@nickpope.me.uk>
|
|
target model changes.
Co-Authored-By: Jacob Walls <38668450+jacobtylerwalls@users.noreply.github.com>
Co-Authored-By: Clifford Gama <cliffygamy@gmail.com>
|
|
Regression in 64b1ac7292c72d3551b2ad70b2a78c8fe4af3249.
|
|
Ordering still depends on pkgutil.iter_modules, which does not guarantee
order, but at least now Django is not introducing additional indeterminism,
causing CircularDependencyError to appear or not appear in some edge cases.
Co-authored-by: Jacob Walls <jacobtylerwalls@gmail.com>
|
|
serialization.
Co-authored-by: Jacob Walls <jacobtylerwalls@gmail.com>
|
|
Thanks Simon Charette for pair programming.
Co-authored-by: Nick Stefan <NickStefan12@gmail.com>
Co-authored-by: Akash Kumar Sen <71623442+Akash-Kumar-Sen@users.noreply.github.com>
Co-authored-by: Simon Charette <charette.s@gmail.com>
|
|
migrations autodetector.
This was creating an unusable Q object solely to call
referenced_base_fields on it.
|
|
GeneratedField.expression.
Thanks to Simon Charette for the suggestion and review.
|
|
their base fields.
Thanks to Colton Saska for the report and to Simon Charette for the review.
|
|
Rewrapped long docstrings and block comments to 79 characters + newline
using script from https://github.com/medmunds/autofix-w505.
|
|
Manually reformatted some long docstrings and comments that would be
damaged by the to-be-applied autofixer script, in cases where editorial
judgment seemed necessary for style or wording changes.
|
|
|
|
migrations.
|
|
|
|
optimizing migrations.
|
|
|
|
keyword arguments.
In Python, keyword arguments must normally be valid identifiers (i.e.,
variable names that follow Python's naming rules). However, Python dicts
can have keys that aren't valid identifiers, like "foo-bar" or "123foo".
This commit ensures that keyword arguments that are nt valid
identifiers, are properly handled when deconstructing an object.
|
|
"Deconstructible" is the spelling that Django has settled on, such as
for `django.utils.deconstruct`. This commit normalizes a
previously-inconsistent class to match the rest of the codebase.
|
|
|
|
|
|
|
|
|
|
|
|
migrations.
|
|
datetime.UTC was added in Python 3.11.
|
|
|
|
|
|
|
|
Co-authored-by: Raphael Gaschignard <raphael@rtpg.co>
|
|
|
|
|
|
This avoids many awkward checks against NOT_PROVIDED and provides symmetry
with Field.has_default() which is also the reason why it wasn't made a
property.
|
|
part of CompositePrimaryKey on SQLite.
|
|
|
|
While AlterUniqueTogether has been documented to be still allowed in historical
migrations for the foreseeable future it has been crashing since 2abf417c815c20
was merged because the latter removed support for Meta.index_together which the
migration framework uses to render models to perform schema changes.
CreateModel(options["unique_together"]) was also affected.
Refs #27236.
Co-authored-by: Simon Charette <charette.s@gmail.com>
|
|
It's not possible for ModelMeta.index_together to exist anymore.
|
|
|
|
|
|
|
|
|
|
_order fields.
Migrations would crash following the removal of an order_with_respect_to
field from a model and the addition of an _order field.
|
|
ManyToManyField.through references on SQLite.
Thank you to dennisvang for the report and Jase Hackman for the test.
Co-authored-by: Jase Hackman <jase.hackman@zapier.com>
|
|
referenced by GeneratedField.expression.
Thank you to Simon Charette for the review.
|
|
|
|
optimizing migrations.
|
|
https://github.com/psf/black/releases/tag/24.1.0
|
|
each operation.
|
