SQL error: 'column reference "id" is ambiguous'

I’ll look into this tomorrow but since the update to 2.2.0 I’ve noticed this issue on sentry;

ProgrammingError at /app/model/1/
column reference "id" is ambiguous
LINE 1: SELECT (CASE WHEN id='1' THEN 0 END) AS "ordering", "...

It complains about validate_values(), as I said I’ll update this ticket later, but it occurs when saving inlines that have autocomplete fields

[alepn]

Hi, the bug ‘column reference “id” is ambiguous’ is resolved. But now we have a new problem when pk name is different from default (id) and the autocomplete field is a foreign key.

The problem is in "SELECT (CASE WHEN the_pk_name=‘877’ THEN 0 END) AS “ordering”, “autht…”.

So I changed the code in models.py FROM:

pk_name = (‘id’ if not getattr(choices.model._meta, ‘pk’, None) else choices.model._meta.pk.name) pk_name = ‘%s.%s’ % (choices.model._meta.db_table, pk_name)

TO:

pk_name = (‘id’ if not getattr(choices.model._meta, ‘pk’, None) else choices.model._meta.pk.attname) pk_name = ‘%s.%s’ % (choices.model._meta.db_table, pk_name)

Then change pk.name to pk.attname it works to me. I think the solution is get the field name in table directly.

[alepn]

I opened a new issue. The problem is get the correct pk_name on subclasses. Take a look #447.