test(mwp-1): Phase 0 — red reproduction of commit-graph gen-0 staleness bug
Reproduces RC-1 (clone-after-push staleness, musehub#106) at both generation sites that share the identical 'gen 0' fallback:
- MWP1_01 (RED): _build_commit_graph_from_raw assigns generation 0 to a commit whose parent is absent from both the mpack and the commit graph, instead of parent_generation + 1. assert 0 == 3. - MWP1_02 (GREEN guard): a genuine root commit (no parents) correctly stays at generation 0 — guards later phases against over-correction. - MWP1_03 (RED): wire_push_unpack_mpack step 9b assigns generation 0 to a newly pushed tip whose parent (the remote tip) has no graph row. Server logs confirm the bug path: 'external_parents=1 found=0' -> 'gen_max=0'.
Also adds the MWP-1 issue body as the local source of truth.
No production code changed — Phase 0 establishes the failing tests only.
Semantic Changes
56 symbols
0 comments
muse hub commit comment sha256:bb4fa850f693e22ecdf6b4f3d12a4996f55972d50184da54b8561d5a52dc2fba --body "your comment"
No comments yet. Be the first to start the discussion.