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 6b03179e126d4df01623dccc162c1579f349e41e 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 2cd0edaa477b327024e4007c8eaf46646dcd0f21 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 ce094e570e0ff085b88b2303e25124331f558e45
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 09fcb70c804b76fccc8fc0ac545873e5ab30c00a.
The admin_widgets tests were issuing click() to the browser but
didn't wait for the effects of those clicks. This caused the resulting
request to be processed concurrently with the test case. When using
in-memory SQLite this caused weird failures.
Also added wait_page_loaded() to admin selenium tests for code
reuse.
Fixed#19856, backpatch of 50677b29af39ca670274fb45087415c883c78b04
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.
Python 2 getpass on Windows doesn't accept unicode, even when
containing only ascii chars.
Thanks Semmel for the report and tests.
Backport of 02e5909f7a from master.
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 f105fbe52b21da206bfbaedf0e92326667d7b2d4 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
1cd6e04cd4f768bcd4385b75de433d497d938f82 for fixing #18676.
Thanks to Dylan Klomparens for the report.
Backpatch of 8ef3235034a1a7616714a5d61486dc68536f74ee
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 0478780b8efdfcbc0d2c17f1459cac45701e1d86
2b5f848207b1dab35afd6f63d0107629c76d4d9a based its changes on #19061
that made the is_active attribute mandatory for user models.
The try/except was not removed for the groups_for_user function.
refs #19780
QuerySet.none() returns an instance of EmptyQuerySet, which may have
undesirable side effects in the presence of custom query set classes.
The implementation of .none() was refactored in master to have the same
effect as .filter(pk__in=[]).
Refs #19652.
Thanks Simon Charrette for the report.
