[blink1073]

INTPYTHON-1082 (https://jira.mongodb.org/browse/INTPYTHON-1082)

Backport of #640.

Changes in this PR

On 6.0.x, an index declared with an ordering expression such as Index(F("priority").desc(), name="...") was silently skipped, and migrate crashed on Index subclasses that report contains_expressions=False (django-tasks-db's ConditionalPartialIndex).

• F() and OrderBy() orderings now translate into MongoDB index keys; anything else, including NULLS FIRST/LAST, is skipped without raising.
• The models.W043 warning now matches whether the index is actually created.
• Dropping a never-created expression index no longer raises IndexNotFound; field-only removals still surface it.

The only delta from #640 is keeping the column_prefix parameter that 6.0.x still needs (main dropped it in the Django 6.1 update), on the rewritten get_pymongo_index_model and on SearchIndex.get_pymongo_index_model.

Test Plan

• New tests cover creation, skipping, warnings, and removal.
• New test exercising the column_prefix parameter (in the second commit).
• On the Django 6.0 fork with a local mongod: the tests fail on unpatched 6.0.x (10 failures, 3 errors), then pass 89 (9 pre-existing skips).
• Django's schema and constraints apps pass; pre-commit run --all-files is clean.

Checklist

Checklist for Author

• Did you update the changelog (if necessary)?
• Is the intention of the code captured in relevant tests?
• If there are new TODOs, has a related JIRA ticket been created?

Checklist for Reviewer

• Does the title of the PR reference a JIRA Ticket?
• Do you fully understand the implementation? (Would you be comfortable explaining how this code works to someone else?)
• Have you checked for spelling & grammar errors?
• Is all relevant documentation (README or docstring) updated?

[Copilot]

🟡 Changes recommended

Unresolved correctness issues affect cross-database checks, expression translation, and embedded-index cleanup.

3 open issues

• Reject sliced F expressions in index translation · New
• Limit MongoDB index checks to MongoDB connections · New
• Preserve embedded model when removing expression indexes · New

What changed in this PR

Backports #640 to the 6.0.x backend, adding MongoDB index support for field-ordering expressions while retaining embedded-field column prefixes.

Changes:

• Translates supported F() and OrderBy() expressions into index keys and aligns warnings with translation support.
• Handles removal of expression indexes skipped by earlier versions.
• Adds regression tests and release notes.

File	Description
tests/​schema_/​test_embedded_model.py	Tests embedded expression indexes with column prefixes.
tests/​indexes_/​test_expressions.py	Tests expression translation, skipping, and removal.
tests/​indexes_/​test_checks.py	Tests expression-index warnings.
tests/​indexes_/​test_base.py	Adds a custom index subclass test fixture.
docs/​releases/​6.0.x.rst	Documents fixes and upgrade guidance.
django_mongodb_backend/​schema.py	Updates expression-index removal handling.
django_mongodb_backend/​indexes.py	Implements expression translation and warning checks.
django_mongodb_backend/​features.py	Clarifies support and adjusts expected test failures.

_{Review effort: Balanced}

[blink1073]

Blocked by #646

[Copilot]

🟡 Changes recommended

An active moderate issue causes embedded-model expression indexes with conditions to generate partial filters at the wrong document path, so the index can match the wrong documents and requires a code fix plus regression coverage.

1 open issue

• Apply embedded prefixes to conditional expression index filters · New

3 resolved since last review

• Preserve embedded model when removing expression indexes
• Limit MongoDB index checks to MongoDB connections
• Reject sliced F expressions in index translation

_{Review effort: Lite}

[Copilot]

🟡 Changes recommended

Unresolved critical and moderate defects can make embedded nullable unique constraints ineffective and cause embedded index creation to fail for list-valued conditions.

2 open issues

• Apply column prefix to nullable unique constraint filters · New
• Handle scalar RHS values when prefixing embedded index conditions · New

1 resolved since last review

• Apply embedded prefixes to conditional expression index filters

_{Review effort: Lite}

[Copilot]

🟢 Approval recommended

No active review issues remain, and the change is supported by regression tests and documented handling for expression indexes, warnings, embedded paths, and index removal.

No open issues 🎉

2 resolved since last review

• Apply column prefix to nullable unique constraint filters
• Handle scalar RHS values when prefixing embedded index conditions

_{Review effort: Lite}

[blink1073]

Closing this as not worth the effort to backport due to the need to handle column_prefix.