There were a couple of regressions related to field pickling. The
regressions were introduced by QuerySet._known_related_objects caching.
The regressions aren't present in master, the fix was likely in
f403653cf1.
Fixed#20157, fixed#20257. Also made QuerySets with model=None
picklable.
A regression caused by d5b93d3281 made .get() error
reporting recurse infinitely on certain rare conditions. Fixed this by
not trying to print the given lookup kwargs.
Backpatch of 266c0bb23e
This reverts commit d1e87eb3ba.
This commit was the cause of a memory leak. See ticket for more details.
Thanks Anssi Kääriäinen for identifying the source of the bug.
Classes overriding __eq__ need a __hash__ such that equal objects have
the same hash.
Thanks akaariai for the report and regebro for the patch.
Backport of e76147a from master.
Backport of 6b03179e12 from master.
Although we're post RC 2, I'm backporting this because it's arguably a
major bug in a new feauture that will prevent several well-known
third-party apps from being ported to Python 3.
When iteration over a queryset raised an exception, the result cache
remained initialized with an empty list, so subsequent iterations returned
an empty list instead of raising an exception
Backport of 2cd0edaa47 from master.
The regression was likely caused by the fix in #19606 which adjusted
Oracle's unicode detection, though it seems this would have been an
issue in some configurations even before.
Backpatch of ce094e570e
Querying the reverse side of nullable to_field relation, where both
sides can contain null values resulted in incorrect results. The reason
was not detecting '' as NULL.
Refs #17541, backpatch of 09fcb70c80.
There was a variable overwrite error in negated join filtering. This
happened when add_filter() was adding the IS NULL condition to the
WHERE clause.
This is not a backport from master as there have been some other
refactorings which made this patch irrelevant.
The patch is from Ian Kelly.
The reason was that the except clause needed to remove a connection
from the django.db.connections dict, but other parts of Django do not
expect this to happen. In addition the except clause was silently
swallowing the exception messages.
Refs #19707, special thanks to Carl Meyer for pointing out that this
approach should be taken.
The change in f105fbe52b made through=None
m2m fields fail in cases where they worked before. It isn't possible to
create such fields using public APIs. The fix is trivial, so it seems
worth adding it.
This is not a backport from master. Master has gotten enough other
changes to related fields internal API that this fix alone isn't enough
to do any good.
When a query had a complex where condition (a condition targeting more
than the base table) a subquery was used for deletion. However, the
query had default ordering from the model's meta and Oracle doesn't
work with ordered subqueries.
The regression was caused by fast-path deletion code introduced in
1cd6e04cd4 for fixing #18676.
Thanks to Dylan Klomparens for the report.
Backpatch of 8ef3235034
PostgreSQL and Oracle always support transactions, so the create table
and rollback test is non-necessary on those DBs.
Thanks to shai for report.
Backpatch of 0478780b8e
The regression was caused by using .none() when querying for related
models, and the origin field's value was None. This resulted in missing
custom related manager subclass as .none() returns plain QuerySet.
This isn't backport from master, in master .none() correctly preserves
the queryset's class.
Patch provided by Simon Charette, with some minor polish by committer.
When &'ing or |'ing querysets, wrong values could be cached, and crashes
could happen.
Thanks Marc Tamlyn for figuring out the problem and writing the patch.
Backport of 07fbc6a.
This is necessary because get_model() checks are case insensitive, and if the swapable check isn't, the
swappable logic gets tied up in knots with models that are partially swapped out.
Thanks to chris@cogdon.org for the report and extensive analysis, and Preston for his work on the draft patch.
Backport of c04c03d from trunk.
Cleaned up the implementation of base convert_values() a little, and
made sure it accepts None as a value for numeric fields.
There are no tests attached. The reason is that not all of the
convert_values() accept None as a value for numeric fields (for example
sqlite3.convert_values()).
The reason the base convert_values() needs to accept None is that this
situation might arise in custom compilers for 3rd party backends. It
is easy to keep the convert_values() working, so lets do that.
Backpatch of 12a96bfa26
SQLite used INSERT INTO tbl SELECT %s UNION SELECT %s, the problem
was that there should have been UNION ALL instead of UNION.
Refs #19351
Backpatch of a27582484c
F() expressions reuse joins like any lookup in a .filter() call -
reuse multijoins generated in the same .filter() call else generate
new joins. Also, lookups can now reuse joins generated by F().
This change is backwards incompatible, but it is required to prevent
dict randomization from generating different queries depending on
.filter() kwarg ordering. The new way is also more consistent in how
joins are reused.
Backpatch of 90b86291d0
The problem is the same as in #10888 which was reintroduced when
bulk_insert was added. Thanks to Jani Tiainen for report, patch and
also testing the final patch on Oracle GIS.
Backpatch of 92d7f541da
Anssi Kääriäinen
Aymeric Augustin
Simon Charette
Gavin Wahl
Marc Tamlyn
Claude Paroz
Grzegorz Nosek
Tim Graham
Jani Tiainen
Russell Keith-Magee
Tai Lee