Skip to content

test: keep invalid-pointer fixtures independent of runtime data - #304

Draft
aberson wants to merge 1 commit into
fix/current-alias-python312from
fix/pointer-fixture-runtime-data
Draft

aberson wants to merge 1 commit into
fix/current-alias-python312from
fix/pointer-fixture-runtime-data

Conversation

@aberson

@aberson aberson commented Sep 28, 2026 •

Copy link
Copy Markdown
Owner

Three invalid/empty current-pointer tests copied the entire live bots/ and src/ trees into temporary directories. In a development checkout those copies included gigabytes of runtime data and caches, making validation depend on local accumulated data.

Use a shared fixture containing the real bots/__init__.py and bots/current/__init__.py plus each test's generated pointer. These imports fail before loading a target bot version. All existing subprocess isolation, failure assertions, and happy-path tests are preserved.

This PR is stacked on #303 so CI includes the Python 3.12 alias repair. Retarget to master after #303 lands.

Validation:

  • Main-checkout Python 3.14: 2025 passed, 2 skipped, 1 deselected in 170.05s, compared with 351.03s in the preceding run of the same checkout (an observed 52% reduction; not a controlled benchmark).
  • Isolated worktree: 2024 passed, 3 skipped, 1 deselected; the additional skip requires the local training database.
  • Linux/Python 3.12 pointer checks: 9 passed.
  • Frontend: 228 passed, 6 skipped; ruff and strict mypy passed. All gate exit codes are 0. Existing formatter wrapping debt outside the changed lines remains unchanged.

aberson added a commit that referenced this pull request Oct 4, 2026
…kfill Issue fields

/repo-sync on documentation/plans/evolve-operational-hardening-plan.md.
Pre-flight: target repo aberson/Alpha4Gate, default branch master, both
prerequisite gates evidenced (plan-review autofix 07017da; plan-wrap passed
2026-09-19; nine redline decisions P-1..P-9 recorded). No cross-plan
collisions — no Phase EH issues existed.

Created 1 umbrella + 10 step issues, all rich bodies (10/10 steps rich on the
richness check). Backfilled all ten **Issue:** fields; repo-sync's verification
grep for blank/TBD values returns empty.

Two corrections made during the sync:
- The plan predicted issue numbering would start at #303, but #303 and #304
  were taken by pull requests in the interim. Actual range is #305 + #306-#315;
  the stale prediction is corrected in the plan and in the umbrella body.
- Step EH.10's issue body initially carried a truncated Depends-on list. The
  plan's §7 line names seven predecessors (EH.1, EH.2, EH.4, EH.5, EH.6, EH.7,
  EH.9); the issue now matches it.

EH.5 is the one frontend-touching step. Its plan Flags line carries no --ui, so
repo-sync added it per the UI-bundle dial and made dashboard evidence part of
acceptance.

Also corrects the Phase EI index row in the master plan: plan-review and
plan-wrap both passed 2026-09-02, but EI has never been redlined, so it needs
/plan-redline before any build. EI is sequenced after EH per EH §6 D-8 — both
edit scripts/evolve.py and must be built one at a time.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant