tools/beads: import_github_golden_test's design-011 golden is stale on main nf-hou ← Beads

closed priority 2 task unassigned

Twinned from https://github.com/cfsh/ninefold/issues/1156 by tools/beads/import_github.py's reconcile pass.

What's broken

tools.beads.import_github_golden_test.ImportGithubGoldenTest.test_real_drift_in_nf_tvc_is_corrected_and_1137_gets_a_new_twin fails on plain origin/main (confirmed via git stash on a branch freshly cut from main — the failure predates and is unrelated to any local change):

``` FAIL: test_real_drift_in_nf_tvc_is_corrected_and_1137_gets_a_new_twin AssertionError: {'nf-tvc': {'comments': [], 'external_ref': No[8370 chars]g.'}} != {'nf-fake1': {'body': "Twinned from https://gi[5864 chars]st'}} Diff is 15526 characters long. Set self.maxDiff to None to see it. : bd's post-apply state does not match test_data/011/bd_children_after.json byte-for-byte — regenerate it (test_data/README.md) if this drift is intentional ```

The test itself is fully mocked (FakeBd/FakeGh fixtures under tools/beads/test_data/011/), not live — so this isn't environment drift, it's import_github.reconcile's actual output no longer matching the checked-in bd_children_after.json golden.

Likely cause

The two most recent commits on main at the time this was found both touch the reconcile/mint path the golden exercises:

Either looks like it could change what body/fields reconcile() writes for a twin, without the golden fixture (tools/beads/test_data/011/bd_children_after.json) being regenerated to match. tools/beads/test_data/README.md documents the regen flow.

Repro

```bash python3 -m unittest tools.beads.import_github_golden_test.ImportGithubGoldenTest.test_real_drift_in_nf_tvc_is_corrected_and_1137_gets_a_new_twin ```

Confirmed on the sandbox box (nf-dev-sg), found incidentally while running ./presubmit.py --all for unrelated work (#1144) — not investigated further, filed unclaimed per CLAUDE.md.

Dependencies

None.

Comments

No comments.

Add a comment