fix(snapshot): stabilize all recorder message ids

This commit is contained in:
Tianyi Cui
2026-07-30 17:51:07 +08:00
parent eb51d4c669
commit 466a2f12c3
12 changed files with 110 additions and 50 deletions
@@ -12,7 +12,7 @@ Message identity needs a weaker structural precondition than aligned records: an
## Decision
Before record or refresh writes fixtures, the suite fingerprints every complete surface message with its top-level `id` removed and groups occurrences across all parent/child logs. It reuses an existing UUID only when one fingerprint resolves to exactly one fresh ID and one existing ID, then applies that mapping to every fresh log. Repeated inherited occurrences with the same ID remain one candidate, while new, changed, duplicate-content, malformed, and conflicting messages keep their fresh IDs.
Before record or refresh writes session fixtures, the shared snapshot support fingerprints every complete surface message with its top-level `id` removed and groups occurrences across all parent/child logs. It reuses an existing UUID only when one fingerprint resolves to exactly one fresh ID and one existing ID, then applies that mapping to every fresh log. Repeated inherited occurrences with the same ID remain one candidate, while new, changed, duplicate-content, malformed, and conflicting messages keep their fresh IDs. ACP, JSON-RPC, TUI, and web recorders pass fixture-ready logs through the same helper before writing.
Refresh write-back uses `normalizeSessionLog` as its volatile-value authority for aligned leaves. It normalizes the original harvested records with the fresh run's ids, cwd, and every cwd alias, while normalizing fixture records with the fixture header context; literal replacements affect only the raw values being written. After existing record alignment, it recursively compares fresh and existing leaves through those normalized records: normalized-equivalent leaves retain the existing raw value, while normalized-distinct leaves retain the fresh semantic value.
@@ -30,6 +30,6 @@ Object fields align by key. Array elements align only when all corresponding arr
## Consequences
Record and refresh no longer rewrite an unchanged unique message UUID solely because another event changed the surrounding record layout. Repeated refreshes also retain aligned fixture values that the normalizer classifies as volatile, and new volatile categories added to the normalizer automatically inherit that write-back behavior. Structural ambiguity remains conservative: unmatched records, conflicting string mappings, resized arrays, strings containing both semantic and volatile changes, and non-unique message fingerprints use fresh values rather than risk reusing misaligned data.
Record and refresh no longer rewrite an unchanged unique message UUID solely because another event changed the surrounding record layout, regardless of whether ACP, JSON-RPC, TUI, or web owns the recording. Repeated refreshes also retain aligned fixture values that the normalizer classifies as volatile, and new volatile categories added to the normalizer automatically inherit that write-back behavior. Structural ambiguity remains conservative: unmatched records, conflicting string mappings, resized arrays, strings containing both semantic and volatile changes, and non-unique message fingerprints use fresh values rather than risk reusing misaligned data.
Focused unit coverage pins scenario-wide parent/child message correlation, unrelated event insertion, record write-back, new/changed/ambiguous messages, recursive object/array behavior, conflicting mappings, fresh cwd aliases, volatile strings, and fresh semantic fields. Keyless refresh coverage proves approval UUIDs, cwd aliases, spill paths, and event-read volatility leave their committed fixtures byte-identical.