Thanks Matt Hegarty for the report and Simon Charette and Natalia Bidart for the reviews.
Regression in f333e35.
Backport of 8b53560eea9f10a1271d3bdf765dc6f969c7d9d5 from main.
A longer term solution is likely to have a better separation of parametrized
DDL altogether to handle checks, constraints, defaults, and generated fields
but such a change would require a significant refactor that isn't suitable
for a backport.
Thanks Adrian Garcia for the report.
Backport of 888b9042b3598bab6557c62de82505eec9ea62ed from main
Bug in 595a2abb58e04caa4d55fb2589bb80fb2a8fdfa1.
Thanks Lucas Lemke Saunitti for the report.
Backport of b98271a6e42107233311d17f5d7bc74fbb47f22c from main
Thanks Brian Ibbotson for the report.
Regression in 7414704e88d73dafbcfbb85f9bc54cb6111439d3.
Backport of 1570ef02f34037d32218d463342592debccf915c from main.
The timezone documentation for Postgres mentions the behavior of
time zone conversion, but links to the wrong setting that controls the
behavior.
Postgres will not return datetimes in the time zone set by the
TIME_ZONE setting, but rather the time zone of the database connection,
which is defined by DATABASES.TIME_ZONE setting falling back to UTC.
This corrects the link in the documentation and adds note that there
are two distinct TIME_ZONE settings and the one most are familiar with
is not considered for PostgreSQL time zone conversion.
Backport of 6e195b800aed7d6a0d2a4c073d8f3507f6beb126 from main
Thanks Simon Alef for the report.
Regression in 868e2fcddae6720d5713924a785339d1665f1bb9.
Backport of a738281265bba5d00711ab62d4d37923764a27eb from main
The note on a possible race condition is inappropriate in this
tutorial setting. To quote Diátaxis:
> Your job is to guide the learner to a successful conclusion. There
> may be many interesting diversions along the way … - ignore them.
Co-Authored-By: Ryan Hiebert <ryan@ryanhiebert.com>
Backport of 0a646c8e08605ba6896ef8f2938231d23c2181cc from main
Regression in 139135627650ed6aaaf4c755b82c3bd43f2b8f51
following deprecation in eedbf930287cb72e9afab1f7208c24b1146b0c4ec.
Backport of a084c5d35a6d00abd261338a374a4424764b4aee from main
Thanks Jon Janzen for the implementation idea.
Thanks Marcus Hoffmann for the report.
Regression in 38e391e95fe5258bc6d2467332dc9cd44ce6ba52.
Backport of d1be05b3e9209fd0787841c71a95819d81061187 from main
Regression in f80669d2f5a5f1db9e9b73ca893fefba34f955e7.
Thanks Sarah Boyce for the review.
Backport of 8db593de05c3516c939b7d4b9eb91e8791f4c79a from main
Thanks Warwick Brown for the report.
Regression in 55519d6cf8998fe4c8f5c8abffc2b10a7c3d14e9.
Backport of 2f14c2cedc9c92373471c1f98a80c81ba299584a from main
Field.db_default accepts either literal Python values or compilables
(as_sql) and wrap the former ones in Value internally.
While 1e38f11 added support for automatic resolving of output fields for
types such as str, int, float, and other unambigous ones it's cannot do
so for all types such as dict or even contrib.postgres and contrib.gis
primitives.
When a literal, non-compilable, value is provided it likely make the
most sense to bind its output field to the field its attached to avoid
forcing the user to provide an explicit `Value(output_field)`.
Thanks David Sanders for the report.
Backport of e67d7d70fa10c06aca36b9057f82054eda45269d from main
MySQL doesn't allow literal DEFAULT values to be used for BLOB, TEXT,
GEOMETRY or JSON columns and requires expression to be used instead.
Regression in 7414704e88d73dafbcfbb85f9bc54cb6111439d3.
Backport of dfc77637ea5c1aa81caa72b1cf900e6931d61b54 from main
