Message8157
The:
test/test_postgresql.py::postgresqlDBTest::testFilteringTransitiveMultilinkSort PASSED [ 63%]
...
test/test_postgresql.py::postgresqlDBTestSchema::testFilteringStringSort PASSED [ 69%]
have one significant difference. The first uses a roundup database and creates the tables
etc. in the database. The database user requires createdb and deletedb privs to run.
The second uses a pre-created database and creates schema's (namespaces) inside of the
database. This database user doesn't need create/delete database privs. Hence the new
roundup_schema user. It places the tables inside the schemas rather than directly in the
database.
(See .github/workflows/ci-test.yml and search for psql for details on the user/db setup.)
I think but am not sure that running a subset of the postgres tests could work
for the schema tests. The differences between these should be in the database user
permissions and database "name".
Maybe we can mark some of the schema tests so a minimal set can run to verify
that we can create a schema and table and test CRUD using the schema.table.
Something like -m "not detail_schema_test" to exclude the marked tests. I'm not
sure marks can be used like this though.
Then we can enable all of the tests for CI to verify that future changes in postgresql,
psycopg2/psycopg3 etc. don't break.
If this goes forward, it should be a new ticket and not buried in this documentation
ticket. |
|
Date |
User |
Action |
Args |
2024-10-22 15:50:40 | rouilj | set | messageid: <1729612240.24.0.398873652473.issue2551366@roundup.psfhosted.org> |
2024-10-22 15:50:40 | rouilj | set | recipients:
+ rouilj, schlatterbeck |
2024-10-22 15:50:40 | rouilj | link | issue2551366 messages |
2024-10-22 15:50:40 | rouilj | create | |
|