The select_related code got confused when it needed to travel a
reverse relation to a model which had different parent than the
originally travelled relation.
Thanks to Trac aliases shauncutts for report and ungenio for original
patch (committed patch is somewhat modified version of that).
Browsers consider the first newline in textareas as some display
artifact, not real content. Hence they are not sending it back to
the server. If we want to keep initial newlines, we have to add one
when we render the textarea.
Thanks bastih for the report and initial patch.
A test in Model Forms test was specifically referring to a fixed
primary key, which was now being used up in a newly committed.
This has been worked around by specifying a higher primary
key.
When calling model_to_dict, improve performance of the generated SQL by
using values_list to determine primary keys of many to many objects. Add
a specific test for this function, test_model_to_dict_many_to_many
Thanks to brian for the original report and suggested fix.
Ensured that special characters passed in to reverse via the
prefix argument are properly escaped so that calls to
django.utils.regex_helpers.normalize and/or string formatting
operations don't result in exceptions.
Thanks to toofishes for the error report.
The dupe avoidance logic was removed as it doesn't seem to do anything,
it is complicated, and it has nearly zero documentation.
The removal of dupe_avoidance allowed for refactoring of both the
implementation and signature of Query.join(). This refactoring cascades
again to some other parts. The most significant of them is the changes
in qs.combine(), and compiler.select_related_descent().
There was a potential data-loss issue involved -- when clearing
instance's m2m assignments it was possible some other instance's
m2m data was deleted instead.
This commit also improved None handling for to_field cases.
Since http://bugs.python.org/issue2193 has been resolved in favour of
the colon in cookie names, we need to test invalid cookie removal using
a different character. "@" is still considered invalid by all sources.
There was a bug introduced in #18676 which caused fast-path deletes
implemented as "DELETE WHERE pk IN <subquery>" to fail if the SELECT
clause contained additional stuff (for example extra() and annotate()).
Thanks to Trac alias pressureman for spotting this regression.
