I've been developing a Django 1.0 project (using Pinax, actaully) on a machine that has PostgreSQL 8.1.11 (Linux), where things work just fine. Now I've deployed to a (still development) customer visible server that has an 8.4 PostgreSQL (also on Linux), which throws errors complaining that equals doesn't know how to compare a character varying with an integer.
Not much analysis reveals that that folks are accustomed to treating the FriendInvitation.status field, a models.CharField with max_length=1 and a set of choices whose DB side representation is a string composed of a single decimal digit, with literal ints (e.g.; self.status = 5 in the FriendInvitation.accept()). This is django_friends version 0.1.2. I'm hitting the problem on a queryset with exclude(status=8). Some testing with psql shows that of: select * from friends_friendshipinvitation where status = '8' ; and select * from friends_friendshipinvitation where status = 8 ; work fine on pg8.1.11, but only the first works on pg8.4 (the column is character varying(1) in both databases). My presumption is that the older PostgreSQL, expecting to have to decide whether unquoted things are strings (e.g.; "status" in the query samples above), used to look at the context and decided that we had meant the 8 as a string. That probably led to difficult to diagnose errors, and they have tightend up in 8.4. I can still get away with not quoting the table and field names (so long as they don't have funny characters), but things that look like numbers now are numbers unless quoted. I can make myself a copy of django_friends that won't get updated when Pinax does, and patch it to work, but since I found my failure in a profiles app that uses friends, I wonder how many other places I'll have to patch. It occurs to me that the ORM could fix this. It knows that status is a CharField. For anything but None (which it should try to store as null) it should be reasonable to apply str() (or unicode()) to the field value before storing it. str(8) -> '8' (and unicode(8) -> u'8'), while str() applied to a string (or unicode() applied to a unicode string) leaves it unchanged (doesn't even copy in, in a few quick tests in python 2.6.2). Then the backend will know to quote it. Thoughts? Workaround suggestions? Bill --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com To unsubscribe from this group, send email to django-users+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/django-users?hl=en -~----------~----~----~----~------~----~------~--~---
