This mostly reverts 6436ec3210,
which was fragile. Instead, if black is present, we use it to format the
expected and actual results, instead of hard-coding the expected
formatted value.
Co-authored-by: Natalia <124304+nessita@users.noreply.github.com>
Thanks Seokchan Yoon for the report, Markus Holtermann for the
triage, and Jake Howard for the review.
Follow-up to CVE-2025-27556 and 39e2297210.
Backport of c880530ddd from main.
Proxy models subclassing a model with a CompositePrimaryKey were
incorrectly reporting check errors because the check that requires only
local fields to be used in a composite pk was evaluated against the proxy
subclass, which has no fields.
To fix this, composite pk field checks are not evaluated against
proxy subclasses, as none of the checks are applicable to proxy
subclasses. This also has the benefit of not double-reporting real check
errors from an invalid superclass pk.
Thanks Clifford Gama for the review.
Backport of 74564946c3 from main.
In Python 3.14, annotations are deferred by default, so we should not
assume that the names in them have been imported unconditionally.
Backport of 6019147229 from main.
tblib 3.2+ makes exception subclasses with __init__() and the default
__reduce__() picklable. This broke the test for
RemoteTestResult._confirm_picklable(), which expects a specific
exception to fail unpickling.
https://github.com/ionelmc/python-tblib/blob/master/CHANGELOG.rst#320-2025-10-21
This fix defines ExceptionThatFailsUnpickling.__reduce__() in a way
that pickle.dumps(obj) succeeds, but pickle.loads(pickle.dumps(obj))
raises TypeError.
Refs #27301. This preserves the intent of the regression test from
52188a5ca6 without skipping it.
Backport of 548209e620 from main.
Migrated `WSGIRequestHandler.log_message()` to use a more robust
`log_message()` helper, which was based of `log_response()` via factoring out
the common bits.
Refs CVE-2025-48432.
Co-authored-by: Natalia <124304+nessita@users.noreply.github.com>
Backport of 9bb83925d6 from main.
GEOSWKTWriter_write() behavior was changed in GEOS 3.12+ to include
parentheses for sub-members (https://github.com/libgeos/geos/pull/903).
MariaDB doesn't accept WKT representations with additional parentheses
for MultiPoint. This is an accepted bug (MDEV-36166) in MariaDB that
should be fixed in the future:
- https://jira.mariadb.org/browse/MDEV-36166
MySQL only supports the ->> when used directly against columns, this can be
inferred by the presence of lhs.output_field.model as model bounds fields are
directly tied to columns.
Purposely don't systematically switch to using JSON_QUOTE(JSON_EXTRACT(...))
as there might be functional indices out there that rely on the SQL remaining
stable between versions.
Thanks Jacob Tavener for the report.
Backport of af84cfba59 from main.
This removes the ability to configure Task enqueueing via a setting,
since the proposed `ENQUEUE_ON_COMMIT` did not support multi-database
setups.
Thanks to Simon Charette for the report.
Follow-up to 4289966d1b.
The logic could likely be adjusted to assign the pre_save value in most cases
to avoid the database transit but it could break in subtle ways so it's not
worth the complexity it would require.
Regression in 94680437a4.
Co-authored-by: Tim Graham <timograham@gmail.com>
Workers created by ParallelTestSuite were not running system
checks in the spawn multiprocessing mode. In general this is
fine, but system checks can have side effects expected by tests.
This patch runs system checks inside of _init_worker, which is
only called by ParallelTestSuite.
This expression automatically escapes its input and allows
fine-grained control over prefix matching and term weighting
via logical combinations.
Thanks Mariusz Felisiak, Adam Zapletal, Paolo Melchiorre,
Jacob Walls, Adam Johnson, and Simon Charette for reviews.
Co-authored-by: joetsoi <joetsoi@users.noreply.github.com>
Co-authored-by: Karl Hobley <karl@kaed.uk>
Co-authored-by: Alexandr Tatarinov <tatarinov1997@gmail.com>
