mirror of
https://github.com/abhigyanpatwari/GitNexus.git
synced 2026-09-09 22:33:39 +00:00
ca54ae35ef
CI caught a bench this branch never ran. `bench/python-scope` is a SEPARATE gate from `bench/scope-capture`; when the receiver-chain work changed capture output I rebaselined the latter's 15 languages and did not know the former existed. My verification loop was two benches; CI runs eight. The drift is intended and has two causes, both from this branch: - `extractMixedChain` now walks THROUGH await and subscript nodes, which is language-agnostic, so Python mints chains where it previously minted none. Probed directly: `(await svc.get_async()).save()` now yields `2|svc|cget_async|a` and `repos[0].save()` yields `2|repos|i`, neither of which existed before. - The codec VERSION prefix moved 1 -> 2, so pre-existing Python chains changed text (`2|svc|cget`). Verified it is those and not something unintended by probing `emitPythonScopeCaptures` for the new name-free step sigils rather than assuming from the fingerprint alone. Checked every other bench CI runs while here — callable-value-flow, scope-emission, python-scope/import-target-fingerprint, cfg, emit-persistence — all PASS. python-scope/measure was the only one. Reproduced locally at the exact CI fingerprint (a0da3e7c…) before touching the baseline, so this is a deterministic capture change and not a CI-environment artifact. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
1 line
65 B
Text
1 line
65 B
Text
a0da3e7c00f603e4bdad91a376b3fc181577a73c2ca1719ab7449d3463c671e0
|