Nested deconstruction should (silently) handle Field.deconstruct() as
well as other arbitrary deconstructable objects. This allows having a
field in the deconstruction of another field.
The first part of django.contrib.postgres, including model and two form
fields for arrays of other data types.
This commit is formed of the following work:
Add shell of postgres app and test handling.
First draft of array fields.
Use recursive deconstruction.
Stop creating classes at lookup time.
Add validation and size parameter.
Add contained_by lookup.
Add SimpleArrayField for forms.
Add SplitArrayField (mainly for admin).
Fix prepare_value for SimpleArrayField.
Stop using MultiValueField and MultiWidget.
They don't play nice with flexible sizes.
Add basics of admin integration.
Missing:
- Tests
- Fully working js
Add reference document for django.contrib.postgres.fields.ArrayField.
Various performance and style tweaks.
Fix internal docs link, formalise code snippets.
Remove the admin code for now.
It needs a better way of handing JS widgets in the admin as a whole
before it is easy to write. In particular there are serious issues
involving DateTimePicker when used in an array.
Add a test for nested array fields with different delimiters.
This will be a documented pattern so having a test for it is useful.
Add docs for SimpleArrayField.
Add docs for SplitArrayField.
Remove admin related code for now.
definition -> description
Fix typo.
Py3 errors.
Avoid using regexes where they're not needed.
Allow passing tuples by the programmer.
Add some more tests for multidimensional arrays.
Also fix slicing as much as it can be fixed.
Simplify SplitArrayWidget's data loading.
If we aren't including the variable size one, we don't need to search
like this.
Previously the FORMAT_MODULE_PATH setting only accepted one string (dotted
module path). A feature has been added to also allow a list of strings.
This is useful when using several reusable third party apps that define new
formats. We can now use them all and we can even override some of the formats
by providing a project-wide format module.
Regression from f51c1f59 when using select_related then prefetch_related
on the reverse side of an O2O:
Author.objects.select_related('bio').prefetch_related('bio__books')
Thanks Aymeric Augustin for the report and tests. Refs #17001.
When using ManifestStaticFilesStorage, deleted static files would be
correctly cleaned up by "collectstatic --clear", but the manifest file
would still contain the stale entries.
Thanks to tedtieken for the report
Considering that these apps only translate their app name, it's not
worth having a whole bunch of po files just for that. The translatable
app name for those apps will be integrated in Django core catalog.
Previously the FORMAT_MODULE_PATH setting only accepted one string (dotted
module path).
This is useful when using several reusable third party apps that define new
formats. We can now use them all and we can even override some of the formats
by providing a project-wide format module.
Loading fixtures were failing since the refactoring in 244e2b71f5 for
inheritance setups where the chain contains abstract models and the
root ancestor contains a M2M relation.
Thanks Stanislas Guerra for the report.
Refs #20946.
Also fixes a slight issue in sqlite3.schema._remake_table where
default values where quoted with "column name" quoting rules.
Reference for quoting: http://www.sqlite.org/lang_expr.html
Thanks Shai Berger for the review. Refs #22424.
Exceptions from the (receiver, exception) tuples returned by
``send_robust()`` now have always their traceback attached as their
``__traceback__`` argument.
While Node class has a useful `__str__`, its `__repr__` is not that
useful. Added a `__repr__` that makes use of the current `__str__`.
This is especially useful since the more popular `Q` class inherits
`tree.Node`. Also created new tests that cover most of `Node` class
functionality.
Added DateTimeCheckMixin to avoid the use of default, auto_now, and
auto_now_add options together. Added the fields.E151 Error that is raised
if one or more of these options are used together.
In django.db.backends.BaseDatabaseWrapper, pulled the creation of
cursors in the non-debug case into a separate method, in order to
make behavior more consistent when overriding the cursor creation
in derived classes.
Fields can now receive the `label_suffix` attribute, which will override
a form's `label_suffix`.
This enhances the possibility to customize form's `label_suffix`, allowing
to use such customizations while using shortcuts such as
`{{ form.as_p }}`.
Note that the field's own customization can be overridden at runtime by
using the `label_prefix` parameter to `BoundField.label_tag()`.
Refs #18134.
Previously, when logging out, the existing session was overwritten by a
new sessionid instead of deleting the session altogether.
This behavior added overhead by creating a new session record in
whichever backend was in use: db, cache, etc.
This extra session is unnecessary at the time since no session data is
meant to be preserved when explicitly logging out.
`sort_dependencies` incorrectly interpreted 'complex' M2M relations
(with explicit through models) as dependencies for a model. This caused
circular complex M2M relations to be unserializable by dumpdata.
Thanks to aneil for the report and outofculture for initial tests.
Previously, known related objects overwrote related objects loaded
though select_related. This could cancel the effect of select_related
when it was used over more than one level.
Thanks boxm for the bug report and timo for bisecting the regression.
Made explicit lack of microsecond handling by built-in datetime form
fields. Used that explicitness to appropriately nix microsecond
values in bound fields. Thanks Claude Paroz for the review.
Compare parameters instead of re.pattern instances, and add the other
parameters to the comparison. Also add a __ne__ to make assertNotEqual
work properly.
In some cases, this could lead to migrations written with Python 2
being incompatible with Python 3.
Thanks Tim Graham for the report and Loïc Bistuer for the advices.
Ordering by reverse foreign key was broken by custom lookups patch
(commit 20bab2cf9d).
Thanks to everybody who helped solving this issue. Special thanks to
Trac alias takis for reporting this.
When custom lookups were added, converting the search lookup to use
the new Lookup infrastructure wasn't done.
Some changes were needed to the added test, main change done by
committer was ensuring the test works on MySQL versions prior to 5.6.
So as the save step is centralized in create(), especially useful
when customizing behavior in subclasses.
Thanks craig.labenz@gmail.com for the report.
Avoided introducing a new regex-based SQL splitter in the migrations
framework, before we're bound by backwards compatibility.
Adapted this change to the legacy "initial SQL data" feature, even
though it's already deprecated, in order to facilitate the transition
to migrations.
sqlparse becomes mandatory for RunSQL on some databases (all but
PostgreSQL). There's no API to provide a single statement and tell
Django not to attempt splitting. Since we have a more robust splitting
implementation, that seems like a good tradeoff. It's easier to add a
new keyword argument later if necessary than to remove one.
Many people contributed to both tickets, thank you all, and especially
Claude for the review.
Refs #22401.
- Fixed bug in get_callable() that caused resolve() to put a string
in ResolverMatch.func.
- Made ResolverMatch.url_name match the actual url name (or None).
- Updated tests that used the string value in ResolverMatch.func, and
added regression tests for this bug.
- Corrected test urls whose dummy view paths caused failures (behavior
that was previously masked by this bug).
Since all state is maintained on the connection at this time and none in
the atomic, it doesn't matter, but it could introduce some subtle bugs
if the implementation changed in the future.
Forward ported code from 1.5 that adds backwards compatibility with legacy message length.
See commit 9e7183073f for details.
Thanks to Ofir Ovadia for the initial patch.
This is useful for tests manually calling migrate inside a testcase,
for normal usage this should make no difference, since there is no
surrounding transaction after all. If there is one we still try to
leave the transaction in a useable state (for postgres at least).
If this commit turns out to be causing issues, settings savepoint=False
is probably the right fix :)
Changed the migration autodetector to remove models last so that FK
and M2M fields will not be left as dangling references. Added a check
in the migration state renderer to error out in the presence of
dangling references instead of leaving them as strings. Fixed a bug
in the sqlite backend to handle the deletion of M2M fields with
"through" models properly (i.e., do nothing successfully).
Thanks to melinath for report, loic for tests and andrewgodwin and
charettes for assistance with architecture.
The warning hint of `_check_test_runner` of 1.6 compatibility had a link
to a general release note. The link should be edited to refer the
relevant "Backwards incompatible changes in 1.6" section that documents
the cause and the possible solutions and workarounds of the warning.
Marc Tamlyn
Thomas Grainger
Ramiro Morales
Martin Brochhaus
Loic Bistuer
Denis Cornehl
Andrew Godwin
Claude Paroz
Erik Romijn
Tim Graham
Markus Amalthea Magnuson
Esau Rodriguez
Florian Apolloner
SaeX
Ramiro Morales
Alex Gaynor
Raphaël Barrois
Baptiste Mispelon
Unai Zalakain
Aymeric Augustin
Shai Berger
Markus Holtermann
Jorge C. Leitão
Stas Rudakou
Vlad Starostin
Russell Keith-Magee
Moayad Mardini
Stephen Burrows
Daniel Pyrathon
mbacho
Tim Martin
Anders Steinlein
Collin Anderson
Julen Ruiz Aizpuru
Matt Robenolt
Rainer Koirikivi
Malcolm Box
David Szotten
Ben Davis
Anssi Kääriäinen
Jakub Roztocil
Chris Beaven
Yehonatan Daniv
Simon Charette
Doug Beck
Bruno Ribeiro da Silva
Ana Krivokapic
Aymeric Augustin
Kevin Christopher Henry
Preston Timmons
Andrew Gorcester
amatellanes