Fixed #26889 -- Fixed missing PostgreSQL index in SchemaEditor.add_field(). #6909
Conversation
|
If we apply the similar reasoning as ticket #25412, we might consider this a data loss issue worthy of backporting to 1.8/1.9. Any thoughts? In that case, I wonder if the schema editor changes might be a bit too invasive for a stable release (concerned about breaking a third-party database backend). |
|
I have created an alternative version of the patch that is less invasive and only involves the PostgreSQL backend. You can view it by viewing the full diff, not just the last commit. If you prefer the 2nd version I will squash the commits. I think I like the original better as it offers a single hook to backends to inject indexes on fields (instead of adding the same index in multiple places). But ultimately both work and I am happy with either. I do see your point wrt backporting the change, so whatever makes that work easier is no problem. Thanks. All feedback welcome. |
|
Looks good. We can merge this approach and then follow up with a refactoring on master for the other approach. I'm not sure the |
| @@ -1908,6 +1908,45 @@ def test_add_textfield_unhashable_default(self): | |||
| editor.add_field(Author, new_field) | |||
|
|
|||
| @unittest.skipUnless(connection.vendor == 'postgresql', "PostgreSQL specific") | |||
| def test_add_field_with_index_with_charfield(self): | |||
There was a problem hiding this comment.
test_add_indexed_charfield seems like a more appropriate name and follows the format used in previous test methods.
|
Made requested changes and squashed commits. Thanks. |
https://code.djangoproject.com/ticket/26889