1a5023883b
Fixed #33257 -- Fixed Case() and ExpressionWrapper() with decimal values on SQLite.
2021-11-08 18:02:56 +01:00
4f8c7fd9d9
Fixed #32980 -- Made models cache related managers.
2021-11-08 08:44:12 +01:00
a934d377af
Fixed #33262 -- Fixed crash of conditional aggregation on Exists().
2021-11-04 06:31:32 +01:00
25157033e9
Fixed #33260 -- Fixed crash when chaining QuerySet.exists() after select_for_update(of=()).
2021-11-03 20:08:14 +01:00
a697424969
Fixed #32996 -- Cached PathInfos on relations.
...
PathInfo values are ostensibly static over the lifetime of the object
for which they're requested, so the data can be memoized, quickly
amortising the cost over the process' duration.
2021-11-03 11:27:04 +01:00
8e3b1cf098
Removed unused model's Options.installed.
...
Unused since f9698c4391
2021-10-28 16:08:59 +02:00
45f48ed4f7
Made F deconstruction omit 'expressions' in the path.
2021-10-21 09:40:52 +02:00
14c8504a37
Refs #25265 -- Allowed Query subclasses to build filters.
2021-10-13 07:31:25 +02:00
2116238d5f
Fixed #33131 -- Improved error messages for clashing reverse accessor names.
2021-10-06 12:33:15 +02:00
0f3e1a54bf
Refs #26430 -- Removed unused branch in sql.Query.get_count().
...
Now that sql.Query.get_aggregation() properly deals with empty result
sets summary Count() annotations cannot result in None.
Unused since 9f3cce172f
2021-10-06 07:25:17 +02:00
903aaa35e5
Fixed #33159 -- Reverted "Fixed #32970 -- Changed WhereNode.clone() to create a shallow copy of children."
...
This reverts commit e441847eca
2021-09-30 11:26:17 +02:00
dd1fa3a31b
Fixed #33018 -- Fixed annotations with empty queryset.
...
Thanks Simon Charette for the review and implementation idea.
2021-09-29 20:23:29 +02:00
ad36a198a1
Fixed #33141 -- Renamed Expression.empty_aggregate_value to empty_result_set_value.
2021-09-29 12:58:01 +02:00
b7fd668b37
Fixed #33033 -- Prevented models.DecimalField from accepting NaN values.
2021-09-28 13:20:14 +02:00
5bac1719a2
Refs #32355 -- Used @functools.lru_cache as a straight decorator.
2021-09-27 09:10:58 +02:00
06c50cee0f
Fixed #33124 -- Avoided accessing the database connections when not necessary.
...
Follow up to bf5abf1bdc
2021-09-24 12:19:39 +02:00
f997c81472
Fixed #33127 -- Added error messages on | and & operators with combined querysets.
2021-09-24 06:59:22 +02:00
3b9fe906bf
Fixed #33133 -- Fixed handling NullBooleanField in historical migrations.
...
Regression in d992f4e3c2
2021-09-23 10:36:30 +02:00
5353e7c250
Refs #27624 -- Optimized Query.clone() for non-combined queries.
...
This avoids constructing a generator expression and a new tuple if the
Query has no combined queries.
2021-09-20 12:34:18 +02:00
e441847eca
Fixed #32970 -- Changed WhereNode.clone() to create a shallow copy of children.
2021-09-20 07:03:10 +02:00
fc91ea1e50
Refs #27624 -- Changed Query.explain_info to namedtuple.
2021-09-17 07:15:12 +02:00
2931d847c2
Optimized Query.clone() a bit.
...
This removes unnecessary "if ... is None" branches, which are already
shallow-copied in the __dict__.copy() call.
2021-09-17 07:12:56 +02:00
306607d5b9
Fixed #32365 -- Made zoneinfo the default timezone implementation.
...
Thanks to Adam Johnson, Aymeric Augustin, David Smith, Mariusz Felisiak, Nick
Pope, and Paul Ganssle for reviews.
2021-09-16 12:11:05 +02:00
0a28b42b15
Fixed #33084 -- Removed incorrect system check for ManyToManyField with limit_choices_to.
2021-09-09 12:07:59 +02:00
691486a5cf
Fixed #33073 -- Fixed queryset crash with aggregation and empty/extra queryset annotation.
2021-09-01 20:59:16 +02:00
bf5abf1bdc
Fixed #33025 -- Avoided accessing the database connections in Query.build_lookup() when not necessary.
...
Of the built-in backends, only Oracle treats empty strings and nulls as
equal, so avoid testing the default connection backend for
interprets_empty_strings_as_nulls if it can be established from the
lookup that it wouldn't affect the lookup instance returned. This
improves performance a small amount for most lookups being built,
because accessing the connections requires touching the thread critical
`Local` which is an expensive operation.
2021-08-19 13:28:59 +02:00
b64db05b9c
Removed unnecessary WhereNode.is_summary.
...
Unnecessary since its introduction in 1df89a60c5
2021-08-09 19:23:29 +02:00
921e4ccb77
Fixed #33003 -- Removed **kwargs from QuerySet._chain().
...
The functionality of updating the __dict__ was only present to allow
for pickling a Prefetch object, which is a comparably rare operation.
Forcing the __getstate__() implementation to handle it allows the
chaining operation to be slightly faster, which is much more common.
2021-08-09 13:28:40 +02:00
5d4f21b16f
Fixed #32983 -- Added system check for redundant related_name on symmetrical M2M fields.
...
Since ManyToManyFields defined with `symmetrical=True` do not add a
related field to the target model, including a `related_name` argument
will never do what the coder likely expects. This makes including
a related_name with a symmetrical model raise a system check warning.
ticket-32983
2021-08-06 11:57:02 +02:00
6f5e07a84d
Refs #32966 -- Refactored out DateTimeCheckMixin._check_if_value_fixed().
2021-07-30 10:15:43 +02:00
6fa5d05dba
Refs #32966 -- Simplified the _check_fix_default_value() implementations.
2021-07-30 09:31:29 +02:00
eebebfe0a9
Refs #32966 -- Added _to_naive() and _get_naive_now() for use in DateTimeCheckMixin classes.
2021-07-30 09:31:24 +02:00
542e749475
Fixed #32966 -- Fixed TimeField.check() crash for timezone-aware times in default when USE_TZ = True.
2021-07-30 09:15:39 +02:00
9662193aea
Refs #32946 -- Changed internal usage of dynamic Q() objects construction to use non-kwargs initialization.
...
This prefers non-kwargs construction of dynamically generated Q()
objects to create a single Q() object instead of many and then
combining them, where possible.
2021-07-28 09:38:46 +02:00
5b8ef8aa5b
Refs #32946 -- Changed Query.add_filter() to take two arguments.
2021-07-28 09:38:42 +02:00
20226fcd46
Fixed #32947 -- Fixed hash() crash on reverse M2M relation when through_fields is a list.
...
Regression in c32d8f33d8
2021-07-26 06:09:29 +02:00
c35b81b864
Fixed #32951 -- Removed Query.where_class & co.
...
Unused since 3caf957ed5
2021-07-22 08:49:20 +02:00
fd999318ad
Fixed #32704 -- Fixed list of deferred fields when chaining QuerySet.defer() after only().
2021-07-20 12:15:29 +02:00
6a970a8b46
Fixed #32944 -- Avoided unnecessary WhereNode.add() calls.
...
Co-authored-by: Mariusz Felisiak <felisiak.mariusz@gmail.com>
2021-07-20 11:59:21 +02:00
501a8db465
Fixed #10929 -- Added default argument to aggregates.
...
Thanks to Simon Charette and Adam Johnson for the reviews.
2021-07-19 13:04:27 +02:00
304f6ff46a
Prevented SQLCompiler.execute_sql() from closing cursor twice.
...
cursor_iter() helper calls cursor.close() in a finally block.
2021-07-16 13:11:46 +02:00
f479df7f8d
Refs #32508 -- Raised Type/ValueError instead of using "assert" in django.db.models.
...
Co-authored-by: Mariusz Felisiak <felisiak.mariusz@gmail.com>
2021-07-15 11:43:33 +02:00
0250340e37
Refs #32074 -- Used Enum.repr() format proposed for Python 3.10.
...
The Python's Steering Council decided to revert changes in the Enum
module (see https://bugs.python.org/issue44559 ) and moved them to
Python 3.11.
Follow up to 5d9b065d3f
2021-07-12 11:46:37 +02:00
f42ccdd835
Fixed #27021 -- Allowed lookup expressions in annotations, aggregations, and QuerySet.filter().
...
Thanks Hannes Ljungberg and Simon Charette for reviews.
Co-authored-by: Mariusz Felisiak <felisiak.mariusz@gmail.com>
2021-07-09 11:43:06 +02:00
5013798fe9
Fixed #32840 -- Optimized Field.get_col().
...
get_col() used "self" as "output_field" when it was not given, and
unnecessarily compared "self" to "self".
Co-authored-by: Chris Jerdonek <chris.jerdonek@gmail.com>
Co-authored-by: Mariusz Felisiak <felisiak.mariusz@gmail.com>
2021-07-07 20:32:57 +02:00
aba9c2de66
Fixed #32226 -- Fixed JSON format of QuerySet.explain() on PostgreSQL.
2021-07-05 09:08:39 +02:00
9f3cce172f
Refs #26430 -- Re-introduced empty aggregation optimization.
...
The introduction of the Expression.empty_aggregate_value interface
allows the compilation stage to enable the EmptyResultSet optimization
if all the aggregates expressions implement it.
This also removes unnecessary RegrCount/Count.convert_value() methods.
Disabling the empty result set aggregation optimization when it wasn't
appropriate prevented None returned for a Count aggregation value.
Thanks Nick Pope for the review.
2021-07-02 07:25:42 +02:00
f3112fde98
Fixed #26430 -- Fixed coalesced aggregation of empty result sets.
...
Disable the EmptyResultSet optimization when performing aggregation as
it might interfere with coalescence.
2021-07-02 06:40:54 +02:00
d8c90d4c22
Fixed #32786 -- Moved subquery ordering clearing optimization to the _in lookup.
...
Co-Authored-By: Simon Charette <charette.s@gmail.com>
2021-06-30 10:08:55 +02:00
053141d31f
Refs #32786 -- Made Query.clear_ordering() not to cause side effects by default.
2021-06-30 08:52:27 +02:00
Matthijs Kooijman
Keryn Knight
Hannes Ljungberg
Mariusz Felisiak
Adam Johnson
Erik Cederstrand
Bernd Wechner
Simon Charette
David Wobrock
Chinmoy Chakraborty
Hasan Ramezani
Carlton Gibson
Nick Touran
Chris Jerdonek
Tom Wojcik
Nick Pope
Daniyal
Ian Foote
Wu Haotian