#33294 closed Bug (invalid)
opclasses missing in PostGIS's 3D geometry indexes
| Reported by: | Tim Graham | Owned by: | nobody |
|---|---|---|---|
| Component: | GIS | Version: | 4.0 |
| Severity: | Normal | Keywords: | |
| Cc: | Triage Stage: | Unreviewed | |
| Has patch: | no | Needs documentation: | no |
| Needs tests: | no | Patch needs improvement: | no |
| Easy pickings: | no | UI/UX: | no |
Description
After 99bc67a9e79256d8a2fcd5742e33a5e79c056539 (refs #32721), the GIST_GEOMETRY_OPS_ND opclass is no longer present for indexes on 3D geometries. You can observe the indexes created while running the test suite and see, for example:
CREATE INDEX "geo3d_city3d_point_id" ON "geo3d_city3d" USING GIST ("point")
instead of
CREATE INDEX "geo3d_city3d_point_id" ON "geo3d_city3d" USING GIST ("point" GIST_GEOMETRY_OPS_ND)
The issue seems to be that when BaseDatabaseSchemaEditor._create_index_sql() calls self._index_columns, it invokes BaseDatabaseSchemaEditor's version (which ignores opclasses) rather than postgresql's.
Attachments (2)
Change History (4)
by , 2 years ago
| Attachment: | pre_99bc67a9e79256d8a2fcd5742e33a5e79c056539 added |
|---|
by , 2 years ago
| Attachment: | post_99bc67a9e79256d8a2fcd5742e33a5e79c056539 added |
|---|
comment:1 by , 2 years ago
| Resolution: | → needsinfo |
|---|---|
| Status: | new → closed |
comment:2 by , 2 years ago
| Resolution: | needsinfo → invalid |
|---|---|
| Severity: | Release blocker → Normal |
Sorry for the false alarm, I should have debugged a bit more carefully. The commit in question merely changed behavior in django-cockroachdb.
Thanks for the report, however it creates exactly the same set of indexes for me 🤔 (see attachments) and it's covered by test_add_3d_field_opclass(). Maybe it's reproducible on a 3rd-party backend.