| Age | Commit message (Collapse) | Author |
|
|
|
Thanks David Smith for reviews.
|
|
This makes the name consistent with the `savepoint_commit` and
`savepoint_rollback` functions. The previous name is maintained as a
deprecated alias.
This also frees up the `savepoint` name, which would allow the context
manager from `django-subatomic` to be included in Django.
Co-authored-by: Lily <code@lilyf.org>
|
|
|
|
action.
|
|
Bug in f30acb184f75fd9260cfd6ddc48a3bbbd49f9c1d. Refs #12090.
|
|
Error pages, admin, and registration templates were updated to use
`{% csp_nonce %}` on their explicit `<script>`, `<link>`, and `<style>`
HTML elements.
Co-authored-by: Antoliny0919 <antoliny0919@gmail.com>
Co-authored-by: Natalia <124304+nessita@users.noreply.github.com>
|
|
on Oracle.
Thanks Simon Charette, JaeHyuck Sa, and Shai Berger for reviews.
|
|
compound queries on SQLite."
This mostly reverts 2314cdf1ff860058a6579bb9f9bac1253fc9ab43,
but keeps the removal of some test skips.
|
|
queries also.
Thanks Shai Berger for the report.
Regression in 087bb9e8f3478d53f12b1737af865992af17c5f2.
(That commit drove more traffic into an error that would have been
reachable only with an explicit order_by() after each union().)
Co-authored-by: Simon Charette <charettes@gmail.com>
Co-authored-by: siddus <dcsid10@gmail.com>
|
|
|
|
Versions of Python prior to 3.15 would incorrectly encode non-ASCII
email addresses using rfc2047, resulting in undeliverable email. The
SMTP EmailBackend detects and prevents that (#35713). Python 3.15 fixes
that behavior (CPython issue gh-122476).
Updated test_rejects_non_ascii_local_part() to feature-detect the fix
(in case it is backported) and check for a representative section of
the Python error message if so; otherwise test for the SMTP EmailBackend
workaround.
Updated comments to clarify need and requirement.
|
|
USE_BLANK_CHOICE_DASH.
Follow-up to 63c56cda133a85a158502891c40465bc0331d3d9.
Modeled on 5d80843ebc5376d00f98bf2a6aadbada4c29365c.
|
|
See DEP 0018.
Added:
* MAILERS setting.
* django.core.mail.mailers dict-like EmailBackend factory.
* `using` argument to mail sending APIs.
* `sent_using` attribute to mail.outbox messages in locmem backend.
* MAILERS in startproject settings template, set to console backend.
* AdminLogHandler.using argument.
* BrokenLinkEmailsMiddleware.send_mail() method.
Updated:
* BaseEmailBackend to track the MAILERS alias used to construct it, and
to report errors for unknown kwargs (OPTIONS).
* EmailBackend implementations to initialize from kwargs (OPTIONS) only
when MAILERS is being used.
* smtp.EmailBackend to require `host` option and to default `port`
option based on SSL/TLS options.
* SimpleTestCase setup to substitute the locmem backend for all defined
MAILERS configurations.
* Django's tests that send mail to define MAILERS.
Deprecated:
* EMAIL_BACKEND and other backend-related EMAIL_* settings.
* mail.get_connection().
* The `connection`, `fail_silently`, `auth_user`, and `auth_password`
arguments to mail functions.
* The EmailMessage.connection attribute.
* BaseEmailBackend support for `fail_silently`. Backends that support
fail_silently (SMTP, console, file) now implement it directly.
* AdminEmailHandler.email_backend argument.
Removed undocumented features without deprecation:
* EmailMessage.get_connection() method. (send() now raises an error if a
subclass has attempted to override it.)
* EmailMessage.send() no longer sets self.connection to the connection
used for sending. (It still _uses_ a pre-existing self.connection.)
* AdminEmailHandler.connection() method. (Init now raises an error if a
subclass has attempted to override it.)
Thanks to Natalia Bidart for shepherding DEP 0018 and for extensive
reviews and suggestions on the implementation.
Thanks to Jacob Rief for the initial implementation and multiple
iterations while refining the design.
Co-authored-by: Jacob Rief <jacob.rief@gmail.com>
|
|
Bug in 019acad112da3d00bb81b3022fd25ccd8457e5d9.
|
|
Thank you to Benjamin Balder Bach and Jacob Walls for reviews.
Co-authored-by: Sarah Boyce <42296566+sarahboyce@users.noreply.github.com>
|
|
CVE-2025-12084 was fixed upstream in CPython and backported to
3.14.2, 3.13.11, and 3.12.13, making this workaround unnecessary.
https://github.com/python/cpython/issues/142145
|
|
|
|
|
|
|
|
Now that calling `QuerySet.select_related()` without arguments is deprecated,
this commit deprecates the corresponding admin features.
|
|
This commit deprecates the "fetch all relations" form of
`QuerySet.select_related()` due to its poor performance characteristics, and
updates several tests relying on that feature to ignore the new warning.
|
|
This commit prepares for the deprecation of `QuerySet.select_related()` with no
arguments by explicitly naming the fields to select in this view.
|
|
This change makes `list_select_related = False` avoid unnecessary joins by only
selecting the related fields that are actually used in `list_display`.
Consequently, it avoids one code path that calls `QuerySet.select_related()`
with no arguments, which is targeted for deprecation.
|
|
|
|
on Python 3.15.
https://github.com/python/cpython/pull/149457
|
|
Replaced the (currently unused) LazySettings._show_deprecation_warning()
with a module-level _show_settings_deprecation_warning() function.
The new function can be called from any settings-related code, not just
LazySettings methods. It correctly distinguishes internal from external
settings usage when override_settings() is involved.
|
|
|
|
Thank you James Walls and James Bligh for reviews.
Co-authored-by: Johannes Maron <johannes@maron.family>
|
|
Schema dependency discovery treated to_field values as raw field names, so attname aliases such as "primary_id" were not matched to the underlying relation field "primary". As a result, AlterField on a unique target field updated direct dependencies but missed transitive attname-based references.
Resolved the dependency matching by comparing resolved remote fields rather than only field names, and updated SQLite's related-table rebuild path to use the same recursive dependency discovery.
Added a regression test covering a transitive relation chain where
ForeignKey(..., to_field="primary_id") must widen along with the unique leaf field it ultimately references.
|
|
Replaced TypeError in `os.path.abspath(None)` with ImproperlyConfigured
error when settings.EMAIL_FILE_PATH is required but missing.
|
|
New default tag `{% csp_nonce_attr %}` was added for explicit CSP nonce
inclusion into `<script>` and `<link>` elements.
`{% csp_nonce_attr %}` renders `nonce="<value>"` when `csp_nonce` is
present in the template context, and renders nothing otherwise.
`{% csp_nonce_attr media %}` renders a `Media` object's assets with the
nonce attr applied to each tag.
Thanks Jacob Walls for the accurate and spot on review comments.
Co-authored-by: Johannes Maron <johannes@maron.family>
|
|
Since the existing user path eventually calls sync_to_async() in acheck_password,
aim for parity with the nonexistent/inactive user branch by adding sync_to_async().
Follow-up to 748ca0a146175c4868ece87f5e845a75416c30e3.
|
|
corresponds to request.META under ASGI.
Because these tests always passed both WSGI environ values and HTTP
headers via `**extra`, this masked a behavior difference between WSGI
and ASGI.
What should happen: everything should be passed via `headers` but for
the default REMOTE_USER case on WSGI, which should be passed via
`**extra`.
Since that was not done, a regression made it into Django 5.2
(50f89ae850f6b4e35819fe725a08c7e579bfd099) where `.header` no longer
corresponded to the request.META key under ASGI. To cope, an ASGI user
would have started(*) sending HTTP headers that match the `.header`
attribute, which may or may not have been edited to remove the HTTP_
prefix. (Note: the default `REMOTE_USER` case did not work under ASGI,
so the change in Django 5.2 had the effect of fixing the default case
but changing the semantic of the custom case.)
(*): Unless they were getting the sync execution path, which didn't have
this bug. See the fix in 0f4fff79d33b7cc84822e66bd1fc16caf8222e3a.
Thanks Mykhailo Havelia and Sarah Boyce for reviews.
|
|
|
|
contains an asterisk.
Thank you Ahmad Sadeddin for the report and Jacob Walls for the review.
|
|
cookie with SESSION_SAVE_EVERY_REQUEST=True.
Thank you Jacob Walls and Natalia Bidart for reviews.
|
|
MemoryFileUploadHandler on ASGI.
In ASGI deployments, Content-Length is not guaranteed to reflect the
actual request body size, so relying on it to gate memory allocation
allowed the limit to be bypassed. The handler now enforces
DATA_UPLOAD_MAX_MEMORY_SIZE regardless of the declared header value.
Thanks to Kyle Agronick for the report. Refs #35289.
Co-authored-by: Natalia <124304+nessita@users.noreply.github.com>
|
|
HttpResponseRedirect.
|
|
This alleviates sync/async duplication.
|
|
We need to switch on whether the request is a WSGI or ASGI request to
know whether to prepend `HTTP_`: we cannot assume sync exceution means
we are running under WSGI, as there could be other sync middleware
forcing sync execution under ASGI.
Thanks Mykhailo Havelia for the report.
|
|
algorithm.
Deprecated the default value of the algorithm argument in
django.utils.crypto.salted_hmac() and django.core.signing.base64_hmac(),
which will change from 'sha1' to 'sha256' in Django 7.0.
|
|
Thanks Eliana Rosselli for the review.
|
|
Co-authored-by: Arfey <Arfey17.mg@gmail.com>
|
|
request.user/auser as handled by login().
Co-authored-by: Arfey <Arfey17.mg@gmail.com>
|
|
handle subclasses.
Co-authored-by: Arfey <Arfey17.mg@gmail.com>
|
|
Setting "check" in OPTIONS["pool"] previously raised TypeError because the
PostgreSQL backend always passed check= to ConnectionPool() and unpacked
**pool_options on top, regardless of CONN_HEALTH_CHECKS. The user's callable
now takes precedence via setdefault(); pool_options is copied first to avoid
mutating the user's settings dict.
|
|
|
|
|
|
Ensure skip_file_prefixes does not match sibling packages like django*.
Bug in f42b89f1bf49a5b89ed852b60f79342320a81c5e
and 34bd3ed944bf38792c631b55e581963d44d52284.
|