Traceback when using locmem cache backend

I get the same one on python 2.4/fc4

Just tried using locmem cache and hit this error too.

--- django/core/cache/backends/locmem.py        (revision 4391)
+++ django/core/cache/backends/locmem.py        (working copy)
@@ -20,7 +20,7 @@
             elif exp < now:
                 should_delete = True
             else:
-                return copy.deepcopy(self._cache[key])
+                return self._cache[key]
         finally:
             self._lock.reader_leaves()
         if should_delete:

Is there a reason for the copy.deepcopy? django.core.cache.backends.simple.CacheClass seems to not use it. The patch above seems to work for me, but I have only tested it using the dev server.

Replying to Gary Wilson <gary.wilson@gmail.com>:

{{{ #!diff --- django/core/cache/backends/locmem.py (revision 4391) +++ django/core/cache/backends/locmem.py (working copy) @@ -20,7 +20,7 @@ elif exp < now: should_delete = True else: - return copy.deepcopy(self._cache[key]) + return self._cache[key] finally: self._lock.reader_leaves() if should_delete: }}} Is there a reason for the copy.deepcopy? django.core.cache.backends.simple.CacheClass seems to not use it. The patch above seems to work for me, but I have only tested it using the dev server.

Patch works for me - Thanks, Gary!

Oliver, "worksforme" isn't there to verify a patch works. Can a dev reopen please?

Replying to SmileyChris:

Oliver, "worksforme" isn't there to verify a patch works. Can a dev reopen please?

Apologies everyone

ticket was reopened.

As Lawrence Oluyede has pointed out on the django-dev list, #599 is the reason deepcopy() is used.

So the locmem cache has to use deepcopy() to avoid memory leaks (items stored in the cache won't be gc'd), but deepcopy() won't work with iterators. I think the answer is to use weakrefs, but I don't really know enough about how they work.

Hi Jacob,

I am afraid weakrefs are not going to help you here. Weakrefs are a means to get a reference to an object without increasing that object's reference count. It's useful when you need to build an intricate data structure with some precise memory management; but accessing an object through a weakref does modify its state, so this is not going to be a workable solution to this here issue.

In truth, this looks like a Python bug; we should probably report it as such, if it hasn't been done yet. There is not much we can do about it on Django's side, however, except perhaps work around it one way or another.

One such way is to cache pickled strings instead of the objects themselves, as in the other caching backends. (The caching API does explicitly mandates that the cached objects be picklable). Here's a patch against trunk that changes the locmem backend's behavior accordingly:

Index: trunk/django/core/cache/backends/locmem.py
===================================================================
--- trunk/django/core/cache/backends/locmem.py  (revision 4664)
+++ trunk/django/core/cache/backends/locmem.py  (working copy)
@@ -2,7 +2,11 @@

 from django.core.cache.backends.simple import CacheClass as SimpleCacheClass
 from django.utils.synch import RWLock
-import copy, time
+import time
+try:
+    import cPickle as pickle
+except ImportError:
+    import pickle

 class CacheClass(SimpleCacheClass):
     def __init__(self, host, params):
@@ -20,7 +24,10 @@
             elif exp < now:
                 should_delete = True
             else:
-                return copy.deepcopy(self._cache[key])
+                try:
+                    return pickle.loads(self._cache[key])
+                except (pickle.PickleError, TypeError):
+                    return default
         finally:
             self._lock.reader_leaves()
         if should_delete:
@@ -35,7 +42,10 @@
     def set(self, key, value, timeout=None):
         self._lock.writer_enters()
         try:
-            SimpleCacheClass.set(self, key, value, timeout)
+            try:
+                SimpleCacheClass.set(self, key, pickle.dumps(value), timeout)
+            except (pickle.PickleError, TypeError):
+                pass
         finally:
             self._lock.writer_leaves()

This actually works surprisingly well. I've not noticed any performance impact.

Hope this helps!

With kind regards,

-- S.

Thanks for the patch Sundance, now all we need are some tests.

Hi Gary,

What kind of test exactly would be needed? By which I mean, isn't the locmem backend already covered by unit tests? What additional tests need I provide? (As you can tell, I'm not very much up to date on unit test practices. :))

Thanks,

-- S.

It appears that the current cache tests only test the the cache backend that is set in the setting file. We should probably change this to test the backend in the settings file along with all the other backends (except memcached). memcached backend would only be tested if it's set in the settings file. These tests might be something worth its own ticket, but at the very least there should be a test included with this ticket that covers the situation that was causing the traceback mentioned in the ticket description.

Hi Gary,

I agree, the testing of all the backends instead of just the currently configured one would be a good idea, worthy of its own ticket.

Here is a patch that adds a test for the problem reported in this here ticket, in the meanwhile. Is this alright? I made sure that this test fails before the bugfix is applied and passes after the bugfix is applied.

Index: tests/regressiontests/cache/tests.py
===================================================================
--- tests/regressiontests/cache/tests.py        (revision 4664)
+++ tests/regressiontests/cache/tests.py        (working copy)
@@ -67,5 +67,14 @@
         time.sleep(2)
         self.assertEqual(cache.get("expire"), None)

+    def test_unserializable_data(self):
+        # Ensure that the cache backend doesn't choke on unserializable
+        # data. (see bug #3012)
+        listiterator = iter([1, 2, 3, 4])
+        defaultvalue = "This is a default value."
+        cache.set("testiterator", listiterator)
+        cached_data = cache.get("testiterator", default=defaultvalue)
+        self.assert_(cached_data in (defaultvalue, listiterator))
+
 if __name__ == '__main__':

(Note that the unit tests already failed on the locmem backend when trying to cache a function (line 57 of tests/regressiontests/cache/tests.py), due to the same issue as above. The bugfix patch submitted earlier also fixes that issue.)

Did I understand correctly what was needed in terms of unit tests, Gary?

With kind regards,

-- S.

Hi Gary, hi Jacob, hi all,

It's been a little while now -- is there something wrong with the bugfix and unit test patches that were herein submitted? Or is this a case of aaaah-not-enough-time-to-keep-track-of-all-tickets, and in that case how can I help?

I'm taking the liberty of removing the needs-tests flag from the ticket, too, I hope that's not presumptuous of me?

Thanks,

-- S.

There is no patch attached to this ticket, so it's not even ready to be moved to "ready for checkin" yet. Please attach a patch for the code, including the tests. Asking us to copy and paste from various comments scattered through the ticket is not a reliable way of reviewing the changes (particularly, since the first attempt is often not the last). Once that is done, we can review it in time, but please have some patience: this is one of many open tickets and they are all treated more or less on an equal basis, with those having larger impact being given slightly higher priority. It will be dealt with in due course.

Hi Malcolm,

Ack, thanks -- I somehow completely missed the 'attach' button when I submitted my bit of code there, so I assumed that code embedding in comments had to be the canonical way to submit a patch, since there was no other obvious way. I've now attached the patch; I hope it's okay now.

I know it must take a lot of time to review all the pending tickets and I'm grateful that you kindly took the time to reply! Don't worry overly much about this particular one, it's only a minor issue. I was just growing worried I had done something wrong -- which, as it turns out, was the case. *coughs*

If there is any other way I can help, please let me know. :)

Thank you for taking the time to point me in the right direction,

-- S.

Saw this problem too (0.96) - the latest patch (locmem-cache-exception-fix.patch) solved the issue; thanks!

This looks good to go - tests and all.

hi any news on this?

It seems that this error is remain in svn

when I use locmem and cache_page, I got the error below:

Traceback (most recent call last):
File "C:\Python25\lib\site-packages\django\core\handlers\base.py" in get_response
  77. response = callback(request, *callback_args, **callback_kwargs)
File "C:\Python25\lib\site-packages\django\utils\decorators.py" in _wrapped_view
  36. result = middleware.process_request(request)
File "C:\Python25\lib\site-packages\django\middleware\cache.py" in process_request
  60. response = cache.get(cache_key, None)
File "C:\Python25\lib\site-packages\django\core\cache\backends\locmem.py" in get
  23. return copy.deepcopy(self._cache[key])
File "C:\Python25\lib\copy.py" in deepcopy
  189. y = _reconstruct(x, rv, 1, memo)
File "C:\Python25\lib\copy.py" in _reconstruct
  337. state = deepcopy(state, memo)
File "C:\Python25\lib\copy.py" in deepcopy
  162. y = copier(x, memo)
File "C:\Python25\lib\copy.py" in _deepcopy_dict
  254. y[deepcopy(key, memo)] = deepcopy(value, memo)
File "C:\Python25\lib\copy.py" in deepcopy
  189. y = _reconstruct(x, rv, 1, memo)
File "C:\Python25\lib\copy.py" in _reconstruct
  322. y = callable(*args)
File "C:\Python25\lib\copy_reg.py" in __newobj__
  92. return cls.__new__(cls, *args)

  TypeError at /
  object.__new__(listiterator) is not safe, use listiterator.__new__()

(In [5703]) Fixed #3012 -- Changed the locmem cache backend to use pickle instead of deepcopy to make it compatible with iterators (which cannot be copied). Patch from Sundance.

(In [5716]) Refs #3012 -- Removed iterator from test_data_types cache test that I added in [5703]. Iterators cannot be pickled either. Left the rest of [5703] there though since it fixed another issue that was causing the test_data_types cache test to fail with the locmem cache backend, the fact that functions cannot be copied.

So after a little bit of playing it seems that the tracebacks above are possibly caused by deepcopy() getting called on an HttpResponse object after iter() had been called on the HttpResponse object. HttpResponse.__iter__ sets self._iterator, and after that a copy() or deepcopy() will fail since self._iterator tries to get copied.

>>> from django.http import HttpResponse
>>> import copy
>>> import pickle
>>> r=HttpResponse()
>>> copy.deepcopy(r)
<django.http.HttpResponse object at 0x83f360c>
>>> pickle.dumps(r)
"ccopy_reg\n_reconstructor\np0\n(cdjango.http\nHttpResponse\np1\nc__builtin__\nobject\np2\nNtp3\nRp4\n(dp5\nS'headers'\np6\n(dp7\nS'Content-Type'\np8\nS'text/html; charset=utf-8'\np9\nssS'_charset'\np10\nS'utf-8'\np11\nsS'_container'\np12\n(lp13\nS''\np14\nasS'cookies'\np15\ng0\n(cCookie\nSimpleCookie\np16\nc__builtin__\ndict\np17\n(dp18\ntp19\nRp20\nsS'_is_string'\np21\nI01\nsb."
>>> iter(r)
<django.http.HttpResponse object at 0x83f344c>
>>> copy.deepcopy(r)
Traceback (most recent call last):
  File "<console>", line 1, in ?
  File "/usr/lib/python2.4/copy.py", line 204, in deepcopy
    y = _reconstruct(x, rv, 1, memo)
  File "/usr/lib/python2.4/copy.py", line 351, in _reconstruct
    state = deepcopy(state, memo)
  File "/usr/lib/python2.4/copy.py", line 174, in deepcopy
    y = copier(x, memo)
  File "/usr/lib/python2.4/copy.py", line 268, in _deepcopy_dict
    y[deepcopy(key, memo)] = deepcopy(value, memo)
  File "/usr/lib/python2.4/copy.py", line 204, in deepcopy
    y = _reconstruct(x, rv, 1, memo)
  File "/usr/lib/python2.4/copy.py", line 336, in _reconstruct
    y = callable(*args)
  File "/usr/lib/python2.4/copy_reg.py", line 92, in __newobj__
    return cls.__new__(cls, *args)
TypeError: object.__new__(listiterator) is not safe, use listiterator.__new__()
>>> pickle.dumps(r)
Traceback (most recent call last):
  File "<console>", line 1, in ?
  File "/usr/lib/python2.4/pickle.py", line 1386, in dumps
    Pickler(file, protocol, bin).dump(obj)
  File "/usr/lib/python2.4/pickle.py", line 231, in dump
    self.save(obj)
  File "/usr/lib/python2.4/pickle.py", line 338, in save
    self.save_reduce(obj=obj, *rv)
  File "/usr/lib/python2.4/pickle.py", line 433, in save_reduce
    save(state)
  File "/usr/lib/python2.4/pickle.py", line 293, in save
    f(self, obj) # Call unbound method with explicit self
  File "/usr/lib/python2.4/pickle.py", line 663, in save_dict
    self._batch_setitems(obj.iteritems())
  File "/usr/lib/python2.4/pickle.py", line 677, in _batch_setitems
    save(v)
  File "/usr/lib/python2.4/pickle.py", line 313, in save
    rv = reduce(self.proto)
  File "/usr/lib/python2.4/copy_reg.py", line 69, in _reduce_ex
    raise TypeError, "can't pickle %s objects" % base.__name__
TypeError: can't pickle listiterator objects

The strange thing is that when testing out the cache_page decorator when using deepcopy() in locmem cache, the exception surfaces, but when using pickle I get no errors.

(In [5828]) Merged revisions 4350-4357,4359-4365,4371-4372,4374-4377,4380-4386,4388,4390-4391,4400-4402,4404-4408,4410,4412-4419,4426-4427,4430-4432,4434,4441,4443-4444,4446-4447,4450,4452-4453,4455-4458,4476,4503,4546,4564-4569,4580-4586,4617,4630,4641-4643,4653-4655,4657,4669,4673-4675,4694-4696,4713-4714,4720-4723,4725-4732,4735-4741,4750,4755,4758,4769-4770,4776-4777,4783-4795,4798,4805-4808,4810,4813-4815,4817,4824,4836,4838-4843,4851-4855,4869,4872,4882-4884,4906,4916,4935-4936,4940-4944,4946-4953,4962-4963,4969,4971-4973,4990,4994-4997,5000-5003,5006-5008,5013-5014,5019-5024,5026-5036,5046-5047,5054-5059,5062,5079,5081-5083,5090,5100-5101,5114,5122-5123,5126,5128,5134-5136,5148-5149,5151,5157,5174,5178,5183-5185,5192-5195,5197-5200,5203,5205,5208,5214,5223-5224,5226-5227,5229-5230,5235-5236,5238-5244,5246-5249,5251,5254-5262,5266-5280,5282-5284,5286,5301,5307,5309-5310,5312,5314-5317,5319-5332,5334-5345,5372-5378,5381-5382,5384,5386,5388-5390,5393-5397,5399-5400,5416,5419-5430,5440-5441,5444-5448,5461-5464,5467,5473-5481,5487-5489,5491-5492,5498-5499,5507-5510,5512,5527,5529,5531-5535,5540-5541,5546,5570,5572-5574,5576-5578,5580-5581,5583-5589,5591,5595-5597,5601-5608,5613,5626-5826 via svnmerge from http://code.djangoproject.com/svn/django/trunk

........

r5626 | russellm | 2007-07-07 10:16:23 +0800 (Sat, 07 Jul 2007) | 2 lines

Added some uncredited authors that worked on the Oracle branch.

........

r5629 | mtredinnick | 2007-07-08 01:15:54 +0800 (Sun, 08 Jul 2007) | 8 lines

Changed HttpRequest.path to be a Unicode object. It has already been URL-decoded by the time we see it anyway, so keeping it as a UTF-8 bytestring was causing unnecessary problems.

Also added handling for non-ASCII URL fragments in feed creation (the portion that was outside the control of the Feed class was messed up).

........

r5630 | mtredinnick | 2007-07-08 02:24:27 +0800 (Sun, 08 Jul 2007) | 4 lines

Fixed #4772 -- Fixed reverse URL creation to work with non-ASCII arguments. Also included a test for non-ASCII strings in URL patterns, although that already worked correctly.

........

r5631 | mtredinnick | 2007-07-08 02:39:23 +0800 (Sun, 08 Jul 2007) | 3 lines

Corrected misleading comment from [5619]. Not sure what I was smoking at the time.

........

r5632 | mtredinnick | 2007-07-08 08:39:32 +0800 (Sun, 08 Jul 2007) | 5 lines

Fixed reverse URL lookup using functions when the original URL pattern was a string. This is now just as fragile as it was prior to [5609], but works in a few cases that people were relying on, apparently.

........

r5636 | mtredinnick | 2007-07-08 19:22:53 +0800 (Sun, 08 Jul 2007) | 4 lines

Fixed #4798-- Made sure that function keyword arguments are strings (for the keywords themselves) when using Unicode URL patterns.

........

r5638 | gwilson | 2007-07-10 10:34:42 +0800 (Tue, 10 Jul 2007) | 2 lines

Fixed #4817 -- Removed leading forward slashes from some urlconf examples in the documentation.

........

r5639 | gwilson | 2007-07-10 10:45:11 +0800 (Tue, 10 Jul 2007) | 2 lines

Fixed #4814 -- Fixed some whitespace issues in tutorial01, thanks John Shaffer.

........

r5640 | gwilson | 2007-07-10 11:26:26 +0800 (Tue, 10 Jul 2007) | 2 lines

Fixed #4812 -- Fixed an octal escape in regular expression that is used in the isValidEmail validator, thanks batchman@free.fr.

........

r5641 | mtredinnick | 2007-07-10 20:02:06 +0800 (Tue, 10 Jul 2007) | 3 lines

Fixed #4823 -- Fixed a Python 2.3 incompatibility from [5636] (it was even demonstrated by existing tests, so I really screwed this up).

........

r5642 | mtredinnick | 2007-07-10 20:03:36 +0800 (Tue, 10 Jul 2007) | 3 lines

Fixed #4804 -- Fixed a problem when validating choice lists with non-ASCII data. Thanks, django@vonposer.de.

........

r5643 | mtredinnick | 2007-07-10 20:33:55 +0800 (Tue, 10 Jul 2007) | 4 lines

Fixed #3760 -- Added the ability to manually set feed- and item-level id elements in Atom feeds. This is fully backwards compatible. Based on a patch from spark343@cs.ubc.ca.

........

r5644 | mtredinnick | 2007-07-11 14:55:12 +0800 (Wed, 11 Jul 2007) | 3 lines

Fixed #4815 -- Fixed decoding of request parameters when the input encoding is not UTF-8. Thanks, Jordan Dimov.

........

r5645 | mtredinnick | 2007-07-11 15:00:27 +0800 (Wed, 11 Jul 2007) | 3 lines

Fixed #4802 -- Updated French translation. Combined contribution from baptiste.goupil@gmail.com and rocherl@club-internet.fr.

........

r5646 | mtredinnick | 2007-07-11 15:12:50 +0800 (Wed, 11 Jul 2007) | 2 lines

Fixed #4753 -- Small update to Spanish translation from Mario Gonzalez.

........

r5649 | jacob | 2007-07-12 08:33:44 +0800 (Thu, 12 Jul 2007) | 1 line

Fixed #4615: corrected reverse URL resolution examples in tutorial 4. Thanks for the patch, simeonf.

........

r5650 | adrian | 2007-07-12 12:43:29 +0800 (Thu, 12 Jul 2007) | 1 line

Added 'New in Django development version' note to docs/syndication_feeds.txt changes from [5643]

........

r5651 | adrian | 2007-07-12 12:44:45 +0800 (Thu, 12 Jul 2007) | 1 line

Edited changes to docs/tutorial04.txt from [5649]

........

r5652 | adrian | 2007-07-12 13:23:47 +0800 (Thu, 12 Jul 2007) | 1 line

Added helpful error message to SiteManager?.get_current() if the user hasn't set SITE_ID

........

r5653 | adrian | 2007-07-12 13:28:04 +0800 (Thu, 12 Jul 2007) | 1 line

Added RequestSite? class to sites framework

........

r5654 | adrian | 2007-07-12 13:29:32 +0800 (Thu, 12 Jul 2007) | 1 line

Improved syndication feed framework to use RequestSite? if the sites framework is not installed -- i.e., the sites framework is no longer required to use the syndication feed framework. This is backwards incompatible if anybody has subclassed Feed and overridden init(), because the second parameter is now expected to be an HttpRequest object instead of request.path

........

r5658 | russellm | 2007-07-12 15:45:35 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4459 -- Added 'raw' argument to save method, to override any pre-save processing, and modified serializers to use a raw-save. This enables serialization of DateFields? with auto_now/auto_now_add. Also modified serializers to invoke save() directly on the model baseclass, to avoid any (potentially order-dependent, data modifying) behavior in a custom save() method.

........

r5659 | russellm | 2007-07-12 19:24:16 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #3770 -- Remove null=True tag from OneToOne? serialization test. OneToOne? fields can't have a value of null.

........

r5660 | russellm | 2007-07-12 19:27:38 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #3768 -- Disabled NullBooleanField? PK serialization test. We can't and don't test null PK values.

........

r5662 | russellm | 2007-07-12 20:33:24 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4837 -- Updated Debian packaging details. Thanks for the suggestion, Yasushi Masuda <whosaysni@gmail.com>.

........

r5663 | russellm | 2007-07-12 20:44:05 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4808 -- Added Chilean regions in localflavor. Thanks, Marijn Vriens <marijn@metronomo.cl>.

........

r5664 | russellm | 2007-07-12 20:48:27 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4745 -- Updated docs to point out that 0 is not a valid SITE_ID when running the tests. Thanks for the suggestion, Lars Stavholm <stava@telcotec.se>.

........

r5665 | russellm | 2007-07-12 20:50:02 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4763 -- Minor typo in cache documentations. Thanks, dan@coffeecode.net.

........

r5666 | russellm | 2007-07-12 20:55:28 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4627 -- Added details on MacPorts? packaging of Django. Thanks, Paul Bissex.

........

r5667 | russellm | 2007-07-12 21:23:11 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4640 -- Fixed import to stringfilter in docs. Proposed solution to move stringfilter into django.template.init introduces a circular import problem.

........

r5668 | russellm | 2007-07-12 21:32:00 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4722 -- Clarified discussion about PYTHONPATH in modpython docs. Thanks for the suggestion, Collin Grady <cgrady@the-magi.us>.

........

r5669 | russellm | 2007-07-12 21:37:59 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4755 -- Modified newforms MultipleChoiceField? to use list comprehension, rather than iteration.

........

r5670 | russellm | 2007-07-12 21:41:27 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4764 -- Added reference to Locale middleware in middleware docs. Thanks, dan@coffeecode.net.

........

r5671 | russellm | 2007-07-12 21:55:19 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4768 -- Converted timesince and dateformat to use explicit floor division (pre-emptive avoidance of Python 3000 compatibility problem), and removed a redundant millisecond check. Thanks, John Shaffer <jshaffer2112@gmail.com>.

........

r5672 | russellm | 2007-07-12 22:00:13 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4775 -- Added some missing Hungarian accents to the urlify.js LATIN_MAP. Thanks, Pistahh <szekeres@iii.hu>.

........

r5673 | russellm | 2007-07-12 22:05:16 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4502 -- Clarified reference to view in tutorial. Thanks for the suggestion, Carl Karsten <carl@personnelware.com>.

........

r5674 | russellm | 2007-07-12 22:11:41 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4522 -- Clarified the allowed filter arguments on the time and date filters. Thanks for the suggestion, admackin@gmail.com.

........

r5675 | russellm | 2007-07-12 22:21:51 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4525 -- Fixed mistaken documentation on arguments to runfcgi. Thanks, Johan Bergstrom <bugs@bergstroem.nu>.

........

r5676 | russellm | 2007-07-12 22:41:32 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4538 -- Split the installation instructions to differentiate between installing a distribution package and installing an official release. Thanks to Carl Karsten for the idea, and Paul Bissex for the patch.

........

r5677 | russellm | 2007-07-12 23:26:37 +0800 (Thu, 12 Jul 2007) | 2 lines

Fixed #4526 -- Modified the test Client login method to fail when a user is inactive. Thanks, marcin@elksoft.pl.

........

r5678 | russellm | 2007-07-13 13:03:33 +0800 (Fri, 13 Jul 2007) | 2 lines

Fixed #3505 -- Added handling for the error raised when the user forgets the comma in a single element tuple when defining AUTHENTICATION_BACKENDS. Thanks for the help identifying this problem, Mario Gonzalez <gonzalemario@gmail.com>.

........

r5679 | mtredinnick | 2007-07-13 16:52:07 +0800 (Fri, 13 Jul 2007) | 3 lines

Fixed #2591 -- Fixed a problem with inspectdb with psycopg2 (only). Patch from Gary Wilson.

........

r5680 | mtredinnick | 2007-07-13 17:09:59 +0800 (Fri, 13 Jul 2007) | 3 lines

Fixed #4807 -- Fixed a couple of corner cases in decimal form input validation. Based on a suggestion from Chriss Moffit.

........

r5681 | mtredinnick | 2007-07-13 17:14:51 +0800 (Fri, 13 Jul 2007) | 3 lines

Fixed #4839 -- Added repr methods to URL classes that show the pattern they contain. Thanks, Thomas G?\195?\188ttler.

........

r5682 | mtredinnick | 2007-07-13 18:56:30 +0800 (Fri, 13 Jul 2007) | 3 lines

Fixed #4842 -- Added slightly more robust error reporting. Thanks, Thomas G?\195?\188ttler.

........

r5683 | mtredinnick | 2007-07-13 19:05:01 +0800 (Fri, 13 Jul 2007) | 3 lines

Fixed #4846 -- Fixed some Python 2.3 encoding problems in the admin interface. Based on a patch from daybreaker12@gmail.com.

........

r5684 | mtredinnick | 2007-07-13 20:03:20 +0800 (Fri, 13 Jul 2007) | 3 lines

Fixed #4861 -- Removed some duplicated logic from the newforms RegexField? by making it a subclass of CharField?. Thanks, Collin Grady.

........

r5685 | mtredinnick | 2007-07-13 21:15:35 +0800 (Fri, 13 Jul 2007) | 3 lines

Fixed #4865 -- Replaced a stray generator comprehension with a list comprehension so that we don't break Python 2.3.

........

r5686 | mtredinnick | 2007-07-13 22:13:35 +0800 (Fri, 13 Jul 2007) | 3 lines

Fixed #4469 -- Added slightly more informative error messages to max- and min-length newform validation. Based on a patch from A. Murat Eren.

........

r5687 | mtredinnick | 2007-07-13 22:14:47 +0800 (Fri, 13 Jul 2007) | 2 lines

Added author credit for [5686]. Refs #4469.

........

r5688 | mtredinnick | 2007-07-13 22:33:46 +0800 (Fri, 13 Jul 2007) | 3 lines

Fixed #4484 -- Fixed APPEND_SLASH handling to handle an empty path value. Thanks, VesselinK.

........

r5689 | mtredinnick | 2007-07-13 22:40:39 +0800 (Fri, 13 Jul 2007) | 2 lines

Fixed #4556 -- Stylistic changes to [5500]. Thanks, glin@seznam.cz.

........

r5690 | gwilson | 2007-07-14 04:36:01 +0800 (Sat, 14 Jul 2007) | 2 lines

Refs #2591 -- Removed int conversion and try/except since the value in the single-item list is already an int. I overlooked this in my original patch, which was applied in [5679].

........

r5691 | adrian | 2007-07-14 05:20:07 +0800 (Sat, 14 Jul 2007) | 1 line

Documented the 'commit' argument to save() methods on forms created via form_for_model() or form_for_instance()

........

r5692 | mtredinnick | 2007-07-14 13:27:22 +0800 (Sat, 14 Jul 2007) | 3 lines

Fixed #4869 -- Added a note that syncdb does not alter existing tables. Thanks, James Bennett.

........

r5693 | mtredinnick | 2007-07-14 20:48:24 +0800 (Sat, 14 Jul 2007) | 3 lines

Fixed #4863 -- Removed comment references to a no-longer present link. Pointed out by Thomas G?\195?\188ttler.

........

r5694 | mtredinnick | 2007-07-14 21:14:28 +0800 (Sat, 14 Jul 2007) | 2 lines

Fixed #4862 -- Fixed invalid Javascript creation in popup windows in admin.

........

r5695 | mtredinnick | 2007-07-14 21:39:41 +0800 (Sat, 14 Jul 2007) | 2 lines

Fixed a problem with translatable strings from [5686].

........

r5696 | mtredinnick | 2007-07-14 22:47:14 +0800 (Sat, 14 Jul 2007) | 3 lines

Fixed #4731 -- Changed management.setup_environ() so that it no longer assumes the settings module is called "settings". Patch from SmileyChris?.

........

r5697 | mtredinnick | 2007-07-14 22:50:35 +0800 (Sat, 14 Jul 2007) | 3 lines

Fixed #4870 -- Removed unneeded import and fixed a docstring in an example. Thanks, Collin Grady.

........

r5698 | adrian | 2007-07-15 00:58:54 +0800 (Sun, 15 Jul 2007) | 1 line

Edited docs/db-api.txt changes from [5658]

........

r5699 | adrian | 2007-07-15 01:04:30 +0800 (Sun, 15 Jul 2007) | 1 line

Negligible capitalization fix in test/client.py docstring

........

r5700 | russellm | 2007-07-15 12:41:59 +0800 (Sun, 15 Jul 2007) | 2 lines

Clarified the documentation on the steps that happen during a save, and how raw save affects those steps.

........

r5701 | gwilson | 2007-07-15 13:03:28 +0800 (Sun, 15 Jul 2007) | 2 lines

Fixed #4310 -- Fixed a regular expression bug in strip_entities function and added tests for several django.utils.html functions. Based on patch from Brian Harring.

........

r5702 | gwilson | 2007-07-15 13:11:06 +0800 (Sun, 15 Jul 2007) | 2 lines

Fixed #4877 -- Fixed typo in testing documentation, patch from John Shaffer.

........

r5703 | gwilson | 2007-07-15 14:24:54 +0800 (Sun, 15 Jul 2007) | 2 lines

Fixed #3012 -- Changed the locmem cache backend to use pickle instead of deepcopy to make it compatible with iterators (which cannot be copied). Patch from Sundance.

........

r5704 | gwilson | 2007-07-15 14:29:45 +0800 (Sun, 15 Jul 2007) | 2 lines

Changed imports to adhere to PEP 8.

........

r5705 | mtredinnick | 2007-07-15 17:39:13 +0800 (Sun, 15 Jul 2007) | 3 lines

Fixed #4880 -- Updated Spanish translation (includes re-encoding to UTF-8). Thanks, Jorge Gajon.

........

r5706 | mtredinnick | 2007-07-15 17:46:42 +0800 (Sun, 15 Jul 2007) | 3 lines

Fixed #4882 -- Updated Argentinean Spanish translation (includes re-encoding to UTF-8). Thanks, Ramiro Morales.

........

r5707 | mtredinnick | 2007-07-15 18:08:05 +0800 (Sun, 15 Jul 2007) | 3 lines

Re-encoded djangojs.po for French and German locales to UTF-8. These were the last two non-UTF-8 PO files.

........

r5708 | mtredinnick | 2007-07-15 18:10:44 +0800 (Sun, 15 Jul 2007) | 6 lines

Fixed #4734 -- Changed message extraction to permit non-ACSII msgid strings. Thanks, krzysiek.pawlik@silvermedia.pl.

This is slightly backwards-incompatible for translators: PO files are now assumed to be in UTF-8 encoding.

........

r5709 | adrian | 2007-07-16 03:34:21 +0800 (Mon, 16 Jul 2007) | 1 line

Edited docs/db-api.txt changes from [5700]

........

r5710 | adrian | 2007-07-16 05:16:32 +0800 (Mon, 16 Jul 2007) | 1 line

Improved docs/templates.txt section on the 'regroup' tag

........

r5711 | mtredinnick | 2007-07-16 11:48:03 +0800 (Mon, 16 Jul 2007) | 2 lines

Updated AUTHORS for [5708].

........

r5712 | mtredinnick | 2007-07-16 11:50:22 +0800 (Mon, 16 Jul 2007) | 3 lines

Fixed #4199 -- Changed date formatting in HTTP expires header to be spec compliant. Thanks, Chris Bennett.

........

r5713 | mtredinnick | 2007-07-16 12:45:45 +0800 (Mon, 16 Jul 2007) | 3 lines

Fixed #4884 -- Fixed an initialisation problem when assigned to settings before accessing them. Thanks, Noam Raphael.

........

r5714 | mtredinnick | 2007-07-16 12:47:52 +0800 (Mon, 16 Jul 2007) | 2 lines

Fixed #4806 -- Updated Simplified Chinese translation. Thanks, limodou.

........

r5715 | mtredinnick | 2007-07-16 12:54:49 +0800 (Mon, 16 Jul 2007) | 3 lines

Fixed #4887 -- Fixed another place where template tag arguments are used directly as function keyword args. Thanks, Brian Rosner.

........

r5716 | gwilson | 2007-07-16 13:00:18 +0800 (Mon, 16 Jul 2007) | 2 lines

Refs #3012 -- Removed iterator from test_data_types cache test that I added in [5703]. Iterators cannot be pickled either. Left the rest of [5703] there though since it fixed another issue that was causing the test_data_types cache test to fail with the locmem cache backend, the fact that functions cannot be copied.

........

r5717 | gwilson | 2007-07-16 13:28:13 +0800 (Mon, 16 Jul 2007) | 2 lines

Cleaned up a couple unused imports and fixed docstrings to follow Python Style Guide.

........

r5718 | mtredinnick | 2007-07-16 17:36:10 +0800 (Mon, 16 Jul 2007) | 3 lines

Fixed #4845 -- Fixed some problems with Unicode usage and caching. Thanks, Jeremy Dunck.

........

r5719 | gwilson | 2007-07-16 21:47:43 +0800 (Mon, 16 Jul 2007) | 2 lines

Removed unused variable and changed comments about permalink decorator into a docstring.

........

r5720 | gwilson | 2007-07-17 06:29:09 +0800 (Tue, 17 Jul 2007) | 2 lines

Fixed #4851 -- Fixed description of an example query in db-api docs.

........

r5721 | mtredinnick | 2007-07-17 12:22:11 +0800 (Tue, 17 Jul 2007) | 2 lines

Fixed #4898 -- Fixed a precendence problem when constructing HTTP Date header.

........

r5722 | mtredinnick | 2007-07-17 18:25:43 +0800 (Tue, 17 Jul 2007) | 3 lines

Fixed #4899 -- Fixed a problem with PO file header generation caused by [5708]. Thanks, Ramiro Morales.

........

r5723 | mtredinnick | 2007-07-19 17:23:45 +0800 (Thu, 19 Jul 2007) | 2 lines

Fixed #4917 -- Updated Swedish translation. Thanks, Pilip Lindborg.

........

r5724 | mtredinnick | 2007-07-19 17:24:36 +0800 (Thu, 19 Jul 2007) | 2 lines

Fixed #3925 -- Added Slovak localflavor items. Thanks, Martin Kos?\195?\173r.

........

r5725 | adrian | 2007-07-20 14:28:56 +0800 (Fri, 20 Jul 2007) | 1 line

Added a db_type() method to the database Field class. This is a hook for calculating the database column type for a given Field. Also converted all management.py CREATE TABLE statements to use db_type(), which made that code cleaner. The Field.get_internal_type() hook still exists, but we should consider removing it at some point, because db_type() is more general. Also added docs -- the beginnings of docs on how to create custom database Field classes. This is backwards-compatible.

........

r5726 | adrian | 2007-07-20 14:34:26 +0800 (Fri, 20 Jul 2007) | 1 line

Simplified the indent level in management.py _get_sql_model_create() by using a 'continue' statement rather than nesting everything in an 'if'

........

r5727 | russellm | 2007-07-20 20:07:58 +0800 (Fri, 20 Jul 2007) | 2 lines

Fixed #4558 -- Modified XML serializer to handle whitespace better around None tags. Thanks to Bill Fenner <fenner@gmail.com> for the report and fix.

........

r5728 | russellm | 2007-07-20 20:15:02 +0800 (Fri, 20 Jul 2007) | 2 lines

Fixed #4897 -- Fixed minor typo in doctest comment.

........

r5729 | russellm | 2007-07-20 21:57:49 +0800 (Fri, 20 Jul 2007) | 2 lines

Fixed #3782 -- Added support for the suite() method recommended by the Python unittest docs. Thanks for the suggestion, rene.puls@repro-mayr.de.

........

r5730 | russellm | 2007-07-20 22:07:54 +0800 (Fri, 20 Jul 2007) | 2 lines

Refs #3782 -- Added documentation note that suite() handling is only in development version.

........

r5731 | russellm | 2007-07-20 22:32:20 +0800 (Fri, 20 Jul 2007) | 2 lines

Fixed #4901 -- Modified assertContains to provide a default check of 'any instances of text in content'. Thanks for the suggestion, nis@superlativ.dk.

........

r5732 | russellm | 2007-07-20 22:42:57 +0800 (Fri, 20 Jul 2007) | 2 lines

Fixed #4738 -- Modified the prompt that is displayed when a test database cannot be created. The existing prompt was misleading if the issue wasn't a pre-existing database. Thanks for the suggestion, John Shaffer <jshaffer2112@gmail.com>.

........

r5733 | adrian | 2007-07-20 23:40:54 +0800 (Fri, 20 Jul 2007) | 1 line

Fixed negligible typo in docstring in tests/regressiontests/test_client_regress/models.py from [5731]

........

r5736 | adrian | 2007-07-21 05:24:30 +0800 (Sat, 21 Jul 2007) | 1 line

Added some additional docs to docs/model-api.txt db_type() section

........

r5738 | russellm | 2007-07-21 11:30:38 +0800 (Sat, 21 Jul 2007) | 2 lines

Fixed #4304 -- Modified sys.exit to os._exit to make sure development server quits when an error occurs attempting to bind to the requested port (e.g., if another server is already running). Thanks, Mario Gonzalez <gonzalemario@gmail.com>.

........

r5739 | russellm | 2007-07-21 12:36:28 +0800 (Sat, 21 Jul 2007) | 2 lines

Minor fix to allow for count=0 in assertContains.

........

r5740 | russellm | 2007-07-21 13:15:19 +0800 (Sat, 21 Jul 2007) | 2 lines

Added test cases for change [5739].

........

r5741 | russellm | 2007-07-21 13:17:20 +0800 (Sat, 21 Jul 2007) | 2 lines

Fixed #4402 -- Modified test client to allow multi-valued inputs on GET requests. Thanks for the suggestion, eddymul@gmail.com.

........

r5743 | gwilson | 2007-07-22 10:18:36 +0800 (Sun, 22 Jul 2007) | 2 lines

Fixed #4945 -- Removed unused GET_ITERATOR_CHUNK_SIZE definition from manager.py. GET_ITERATOR_CHUNK_SIZE is already defined in query.py. Thanks zigiDev@mac.com.

........

r5744 | gwilson | 2007-07-22 11:09:24 +0800 (Sun, 22 Jul 2007) | 2 lines

Added docstrings to shortcuts module and functions.

........

r5745 | gwilson | 2007-07-22 11:12:50 +0800 (Sun, 22 Jul 2007) | 2 lines

Shortcut functions do not accept QuerySet objects, yet :)

........

r5746 | gwilson | 2007-07-22 11:41:11 +0800 (Sun, 22 Jul 2007) | 2 lines

Fixed #4373 -- Modified the get_object_or_404/get_list_or_404 shortcuts to also accept QuerySets. Thanks SuperJared?.

........

r5747 | gwilson | 2007-07-22 11:45:03 +0800 (Sun, 22 Jul 2007) | 2 lines

Corrected typo in [5746].

........

r5750 | gwilson | 2007-07-23 12:45:01 +0800 (Mon, 23 Jul 2007) | 2 lines

Fixed #4952 -- Fixed the get_template_sources functions of the app_directories and filesystem template loaders to not return paths outside of given template directories. Both functions now make use of a new safe_join utility function. Thanks to SmileyChris? for help with the patch.

........

r5752 | russellm | 2007-07-23 20:14:32 +0800 (Mon, 23 Jul 2007) | 2 lines

Fixed #3771 -- Modified the test runner to observe the --noinput argument controlling script interactivity. This means that test scripts can now be put in a buildbot environment. This is a backwards incompatible change for anyone that has written a custom test runner. Thanks for the suggestion, moof@metamoof.net.

........

r5753 | russellm | 2007-07-23 21:52:59 +0800 (Mon, 23 Jul 2007) | 2 lines

Added documentation for a test runner argument that has always been present, but was undocumented.

........

r5756 | adrian | 2007-07-25 11:12:31 +0800 (Wed, 25 Jul 2007) | 1 line

Changed docstring additions from [5744] to use active verbs ('returns' instead of 'return')

........

r5757 | adrian | 2007-07-25 11:15:05 +0800 (Wed, 25 Jul 2007) | 1 line

Added 'New in Django development version' to docs/db-api.txt change from [5746]

........

r5758 | adrian | 2007-07-25 11:18:17 +0800 (Wed, 25 Jul 2007) | 1 line

Changed safe_join() docstring from [5750] to use active verbs. See also [5756]

........

r5764 | gwilson | 2007-07-26 13:01:53 +0800 (Thu, 26 Jul 2007) | 2 lines

Fixed #4971 -- Fixed some escaping and quoting problems in the databrowse contrib app. Based on patch from Johann Queuniet.

........

r5765 | adrian | 2007-07-27 01:16:34 +0800 (Fri, 27 Jul 2007) | 1 line

Added section to docs/contributing.txt about docstring coding style

........

r5766 | mtredinnick | 2007-07-27 06:59:34 +0800 (Fri, 27 Jul 2007) | 2 lines

Added support for database cache table in test database.

........

r5767 | adrian | 2007-07-28 05:53:02 +0800 (Sat, 28 Jul 2007) | 1 line

Added unit test that confirms a bug in ValuesQuerySets? that have extra(select) specified. If the select dictionary has several fields, Django assigns the wrong values to the select-field names

........

r5768 | adrian | 2007-07-28 06:07:42 +0800 (Sat, 28 Jul 2007) | 1 line

Fixed bug with using values() and extra(select) in the same QuerySet?, with a select dictionary containing more than a few elements. This bug was identified in unit test from [5767]. The problem was that we were relying on the dictionary's .items() ordering, which is undefined

........

r5769 | russellm | 2007-07-28 12:02:52 +0800 (Sat, 28 Jul 2007) | 2 lines

Fixed #4460 -- Added the ability to be more specific in the test cases that are executed. This is a backwards incompatible change for any user with a custom test runner. See the wiki for details.

........

r5770 | russellm | 2007-07-28 15:27:53 +0800 (Sat, 28 Jul 2007) | 2 lines

Fixed #4995 -- Fixed some problems in documentation ReST formatting. Thanks, Simon G.

........

r5771 | simon | 2007-07-29 02:30:40 +0800 (Sun, 29 Jul 2007) | 1 line

After discussing with Malcolm, added set_unusable_password() and has_usable_password() methods to the User object, plus tests and updated documentation

........

r5774 | adrian | 2007-07-30 02:21:16 +0800 (Mon, 30 Jul 2007) | 1 line

Added 'New in Django development version' to changes in docs/authentication.txt from [5771]

........

r5778 | gwilson | 2007-07-31 01:25:35 +0800 (Tue, 31 Jul 2007) | 4 lines

Fixed call to ugettext, which is imported as _. Changed raise to conform to PEP 3109 and wrapped the long line. Added beginnings of tests for model fields.

........

r5782 | gwilson | 2007-08-01 13:41:32 +0800 (Wed, 01 Aug 2007) | 2 lines

Fixed #4228 -- Removed hardcoding of RadioFieldRenderer in the RadioSelect Widget so that the display of RadioSelects can be more easily customized. BoundField.__unicode__ also no longer special cases RadioSelect since RadioSelect.render() now returns a string like every other Widget.

........

r5783 | gwilson | 2007-08-01 13:52:18 +0800 (Wed, 01 Aug 2007) | 2 lines

Fixed #5037 -- Fixed use of wrong field type in a db-api docs example, thanks ubernostrum.

........

r5796 | gwilson | 2007-08-04 11:19:14 +0800 (Sat, 04 Aug 2007) | 2 lines

Fixed #5078 -- Fixed several broken links to the syndication documentation.

........

r5797 | gwilson | 2007-08-04 11:36:58 +0800 (Sat, 04 Aug 2007) | 2 lines

Changed the 0.95 release notes to point to the 0.95 documentation index.

........

r5798 | gwilson | 2007-08-04 11:39:24 +0800 (Sat, 04 Aug 2007) | 2 lines

Changed several documentation links to be relative.

........

r5799 | gwilson | 2007-08-04 22:41:49 +0800 (Sat, 04 Aug 2007) | 2 lines

Refs #3397 -- Corrected the Exception that is caught when ordering by non-fields (added in [4596]), thanks glin@seznam.cz.

........

r5800 | gwilson | 2007-08-04 22:52:13 +0800 (Sat, 04 Aug 2007) | 2 lines

Fixed #5083 -- Fixed typo in newforms documentation, thanks Rik.

........

r5801 | gwilson | 2007-08-05 12:39:52 +0800 (Sun, 05 Aug 2007) | 2 lines

Refs #5089 -- Added file name to poll detail template examples in the tutorial.

........

r5802 | gwilson | 2007-08-05 12:42:26 +0800 (Sun, 05 Aug 2007) | 2 lines

Changed some more links to be relative in the documentation. I had a couple unsaved files that didn't get in with [5798].

........

r5803 | gwilson | 2007-08-05 13:14:46 +0800 (Sun, 05 Aug 2007) | 2 lines

Fixed #2101 -- Renamed maxlength argument to max_length for oldforms FormFields and db model Fields. This is fully backwards compatible at the moment since the legacy maxlength argument is still supported. Using maxlength will, however, issue a PendingDeprecationWarning when used.

........

r5804 | russellm | 2007-08-05 15:39:36 +0800 (Sun, 05 Aug 2007) | 2 lines

Fixed #4001 -- Added dynamic save_m2m method() to forms created with form_for_model and form_for_instance on save(commit=False).

........

r5807 | adrian | 2007-08-06 12:36:43 +0800 (Mon, 06 Aug 2007) | 1 line

Fixed #5074 -- Added link to audio clip of 'Django' pronunciation

........

r5808 | adrian | 2007-08-06 12:52:14 +0800 (Mon, 06 Aug 2007) | 1 line

Edited docs/newforms.txt changes from [5804]

........

r5809 | adrian | 2007-08-06 13:04:27 +0800 (Mon, 06 Aug 2007) | 1 line

Fixed #5082 -- Enabled tab completion in 'django-admin.py shell' for objects that were imported into the global namespace at runtime. Thanks, dusk@woofle.net

........

r5810 | adrian | 2007-08-06 13:06:15 +0800 (Mon, 06 Aug 2007) | 1 line

Fixed #5077 -- django/utils/encoding.py no longer imports settings, as it doesn't use that module. Thanks, Collin Grady

........

r5811 | adrian | 2007-08-06 13:07:38 +0800 (Mon, 06 Aug 2007) | 1 line

Fixed #5071 -- Fixed 'global name ugettext is not defined' error in django.core.validators. Thanks, Marco Bonetti

........

r5812 | adrian | 2007-08-06 13:13:06 +0800 (Mon, 06 Aug 2007) | 1 line

Fixed #5064 -- Fixed potentially confusing sentence in docs/authentication.txt. Thanks, Collin Grady

........

r5813 | adrian | 2007-08-06 13:16:35 +0800 (Mon, 06 Aug 2007) | 1 line

Fixed #5053 -- Added 'action' attribute to <form> tags that didn't have that attribute in docs/newforms.txt examples. Perfectionism appreciated, trickyb

........

r5814 | adrian | 2007-08-06 13:27:58 +0800 (Mon, 06 Aug 2007) | 1 line

Added a closing </p>' to a code example in docs/email.txt

........

r5815 | adrian | 2007-08-06 13:28:45 +0800 (Mon, 06 Aug 2007) | 1 line

Fixed #5006 -- Fixed incorrect/outdated docstring for the 'if' template tag. Thanks, Thomas Petazzoni

........

r5816 | adrian | 2007-08-06 13:33:18 +0800 (Mon, 06 Aug 2007) | 1 line

Added note to docs/model-api.txt about help_text not being escaped in the admin interface

........

r5817 | adrian | 2007-08-06 13:34:45 +0800 (Mon, 06 Aug 2007) | 1 line

Fixed #4985 -- Clarified location of HttpResponse in docs/request_response.txt. Thanks for raising the issue, rainer.mansfeld@romulo.de

........

r5818 | adrian | 2007-08-06 13:37:17 +0800 (Mon, 06 Aug 2007) | 1 line

Fixed #4980 -- Removed 'forms' from the 'not considered stable and will be rewritten' section of docs/api_stability.txt. They've already been rewritten.

........

r5819 | russellm | 2007-08-06 21:58:56 +0800 (Mon, 06 Aug 2007) | 2 lines

Fixed #3297 -- Implemented FileField? and ImageField? for newforms. Thanks to the many users that contributed to and tested this patch.

........

r5820 | russellm | 2007-08-06 22:17:10 +0800 (Mon, 06 Aug 2007) | 2 lines

Added note that FileField? and ImageField? are only in development version. There are also some minor backwards compatibility issues with the changes introduced in [5819] - see the wiki for details.

........

r5823 | adrian | 2007-08-07 04:27:04 +0800 (Tue, 07 Aug 2007) | 1 line

Fixed British spelling of 'customize' and 'behavior' in Manager.get_query_set() docstring

........

r5824 | adrian | 2007-08-07 10:18:36 +0800 (Tue, 07 Aug 2007) | 1 line

Fixed #5105 -- Fixed two ReST errors in docs/newforms.txt. Thanks, Ramiro Morales

........

r5825 | adrian | 2007-08-07 10:33:11 +0800 (Tue, 07 Aug 2007) | 1 line

Fixed #5097 -- Made various updates and corrections to the documentation. Thanks, Nicola Larosa

........

r5826 | russellm | 2007-08-07 19:20:15 +0800 (Tue, 07 Aug 2007) | 2 lines

Removed a redundant directory join during FileField? form saving. Thanks to David Danier's eagle eyes for picking up this one.

........