20101103 – worked on ticket #2¶
Weiter an docs/tickets/2.
Upps, da hatte ich mir einen neueren Patch gefragt, und jetzt ist der auch schon wieder veraltet:
L:\snapshots\django>patch --dry-run -p0 < 7539.on_delete.r14291.diff
patching file `django/db/models/sql/subqueries.py'
patching file `django/db/models/base.py'
patching file `django/db/models/options.py'
patching file `django/db/models/fields/related.py'
patching file `django/db/models/__init__.py'
patching file `django/db/models/deletion.py'
patching file `django/db/models/query.py'
patching file `django/db/models/query_utils.py'
patching file `django/core/management/validation.py'
patching file `django/contrib/admin/util.py'
Hunk #2 succeeded at 92 with fuzz 2 (offset -13 lines).
Hunk #3 succeeded at 117 (offset -1 lines).
Hunk #4 succeeded at 143 (offset -13 lines).
patching file `tests/modeltests/invalid_models/models.py'
Hunk #1 succeeded at 210 with fuzz 2 (offset 3 lines).
Hunk #2 FAILED at 321.
1 out of 2 hunks FAILED -- saving rejects to tests/modeltests/invalid_models/models.py.rej
patching file `tests/modeltests/delete/tests.py'
patching file `tests/modeltests/on_delete/__init__.py'
patching file `tests/modeltests/on_delete/models.py'
patching file `tests/regressiontests/admin_util/tests.py'
patching file `tests/regressiontests/admin_util/models.py'
Aber sowieso ist Carl Meyer inzwischen daran am arbeiten.
Ob ich seinen work in progress mal testen sollte? Dazu muss ich mir zunächst Git installieren und Carls Version von Django runterladen:
L:\snapshots\carljm>git clone git://github.com/carljm/django.git
Cloning into django...
remote: Counting objects: 82115, done.
remote: Compressing objects: 100% (18456/18456), done.
remote: Total 82115 (delta 62404), reused 81577 (delta 61973)
Receiving objects: 100% (82115/82115), 29.03 MiB | 115 KiB/s, done.
Resolving deltas: 100% (62404/62404), done.
In meiner sitecustomize.py:
#~ site.addsitedir(r"l:\snapshots\django")
site.addsitedir(r"l:\snapshots\carljm\django")
Dann füge ich ein on_delete=RESTRICT in notes.Note.type
hinzu:
type = models.ForeignKey(NoteType,
blank=True,null=True,
verbose_name=_('Note type'),
on_delete=RESTRICT)
Und… nichts funktioniert! Django verhält sich genau wie vorher. Ich schätze, dass ich mir aus Carls Git-Hub lediglich den Master runtergeladen habe, statt den Code, an dem er eigentlich gearbeitet hat. Denn in meiner Kopie gibt es keine Datei, die den String on_delete enthält.
Und die Dokumentation von git pull und `git merge hilft mir nicht. Versuche:
L:\snapshots\carljm\django>git checkout t7539
Switched to branch 't7539'
L:\snapshots\carljm\django>git pull
You asked me to pull without telling me which branch you
want to merge with, and 'branch.t7539.merge' in
your configuration file does not tell me, either. Please
specify which branch you want to use on the command line and
try again (e.g. 'git pull <repository> <refspec>').
See git-pull(1) for details.
If you often merge with the same branch, you may want to
use something like the following in your configuration file:
[branch "t7539"]
remote = <nickname>
merge = <remote-ref>
[remote "<nickname>"]
url = <url>
fetch = <refspec>
See git-config(1) for details.
Also momentan bin ich einfach noch zu doof für Git, und mein Ticket #2 muss warten, basta.
Zwischendurch habe ich auch nochmal probiert, meine Django-Testsuite zu benutzen. Siehe auch https://www.lino-framework.org/admin/django_tests.html.
Ich habe mir in L:\snapshots\django\tests
eine runtests.bat
mit folgendem Inhalt gemacht:
python runtests.py --settings=test_sqlite %*
Aber wenn ich damit meine Django-Kopie teste, dann sagt er mir hunderte Failures:
........................................................................................................................
........................................................................................................................
...........................s............................................F...............................................
........................................................................................................................
..................FF...............................................F....................................................
....................................x..............................................................................s....
.........F..............................................................................................................
........................................................................................................................
..........................................................s..........FF.......EEEEEFFFF.................F.FFF.E.........
......................l:\snapshots\carljm\django\django\template\defaulttags.py:52: UserWarning: A {% csrf_token %} was
used in a template, but the context did not provide the value. This is usually caused by not using RequestContext.
warnings.warn("A {% csrf_token %} was used in a template, but the context did not provide the value. This is usually
caused by not using RequestContext.")
................
Um irgendwann zu enden:
======================================================================
FAIL: testFlagPost (regressiontests.comment_tests.tests.moderation_view_tests.FlagViewTests)
POST the flag view: actually flag the view (nice for XHR)
----------------------------------------------------------------------
Traceback (most recent call last):
File "L:\snapshots\django\tests\regressiontests\comment_tests\tests\moderation_view_tests.py", line 23, in testFlagPot
self.assertEqual(response["Location"], "http://testserver/flagged/?c=%d" % pk)
AssertionError: 'http://testserver/accounts/login/?next=/flag/1/' != 'http://testserver/flagged/?c=1'
======================================================================
FAIL: testFlagPostTwice (regressiontests.comment_tests.tests.moderation_view_tests.FlagViewTests)
Users don't get to flag comments more than once.
----------------------------------------------------------------------
Traceback (most recent call last):
File "L:\snapshots\django\tests\regressiontests\comment_tests\tests\moderation_view_tests.py", line 30, in testFlagPotTwice
c = self.testFlagPost()
File "L:\snapshots\django\tests\regressiontests\comment_tests\tests\moderation_view_tests.py", line 23, in testFlagPot
self.assertEqual(response["Location"], "http://testserver/flagged/?c=%d" % pk)
AssertionError: 'http://testserver/accounts/login/?next=/flag/1/' != 'http://testserver/flagged/?c=1'
======================================================================
FAIL: testFlagSignals (regressiontests.comment_tests.tests.moderation_view_tests.FlagViewTests)
Test signals emitted by the comment flag view
----------------------------------------------------------------------
Traceback (most recent call last):
File "L:\snapshots\django\tests\regressiontests\comment_tests\tests\moderation_view_tests.py", line 64, in testFlagSinals
self.testFlagPost()
File "L:\snapshots\django\tests\regressiontests\comment_tests\tests\moderation_view_tests.py", line 23, in testFlagPot
self.assertEqual(response["Location"], "http://testserver/flagged/?c=%d" % pk)
AssertionError: 'http://testserver/accounts/login/?next=/flag/1/' != 'http://testserver/flagged/?c=1'
Variante:
runtests.bat --noinput 2> 20101103.log
Creating test database 'default'...
Creating test database 'other'...
Destroying old test database 'other'...
Okay, diese Log-Datei könnte ich mir bei Gelegenheit mal vornehmen.
TODO: Django Test-Suite ans Laufen kriegen und Git-Benutzung lernen, um bei Diskussionen zu Django-Tickets mitreden zu können.
Eine erste Lösungsidee war, dass ich manuell in jedem Fall eine eigene delete-Methode schreibe:
def delete(self):
if self.note_set.count() > 0:
raise IntegrityError("Must delete all Note objects before deleting NoteType")
super(NoteType, self).delete()
Also zumindest mit NoteType wird die Panne nicht mehr passieren. Das Gleiche müsste ich noch für viele andere Fälle machen. Aber das ist natürlich viel Tipperei im Vergleich zu einem on_delete=RESTRICT.
Another problem that must be addressed even if we had on_delete=RESTRICT already working:
When a user clicks on the Delete button of a NoteType for which Notes exist, Lino will still ask a Confirmation “Delete 1 rows. Are you sure?”. Only when the user confirms, Lino will say that it isn’t allowed to delete this record.
This is not optimal. Best would be to disable the Delete button, at least in Detail view. Which means that the server should return this information together with the record.
The Delete button in a grid view cannot be disabled/enabled à priory since there may be several rows selected. The grid would need an URI request where it can ask whether it is allowed to delete an object, without actually deleting it… no: if the UI wants to handle this case, it can simply ask for the detailed information of the record in question.
Temporary solution:
lino.reports.Report.disable_delete
and
for example notes.NoteType.disable_delete()
:
def disable_delete(self,request):
if self.note_set.count() > 0:
return _("Must delete all Note objects before deleting NoteType")