Repository navigation
fix(reporting): align source timezone days to local midnight - #1926
rachitm022 wants to merge 1 commit into
Conversation
There was a problem hiding this comment.
Ladon verdict: Approve
Approve.
Clean timezone-alignment fix. Anchors source_timezone daily periods at local midnight (naive datetime(1970,1,1) in _schedule_clock) and routes consumer-status validation through the same clock, eliminating the prior epoch-UTC phase that misaligned local days. The source_timezone alignment joins an already-existing wire vocabulary (_ALIGNMENTS, resolution.py, offerings.py referenced it at base), so this is additive under fix: — no semver signal issue and no new version-translation pathway owed. Production admission correctly relaxes zone verification only when the source offering is account-resolved (source_timezone is None); fixed offerings still reject a mismatched config zone (BINDING_MISMATCH). Test coverage is strong (DST spring/fall, Kolkata half-hour offset, period-key retry identity, unchanged-phase regression guard, requires-zone/no-anchor negative).
No critical/high/medium findings. One Low (bare ValueError ordering before LedgerConflictError, dead for already-admitted configs) is omitted. high_risk false, gated_paths false, no author team gate. Falls through to row 9.
Why
Part of AI-22388's calendar prerequisite: source-timezone daily reporting must follow the source's civil day, including DST transitions. This change does not provide package-grain full coverage.
What changed
Validation
make ci-localwith six pytest workers: 22,075 passed, 3,031 skipped, seven Mac-only failures in process cleanup and Linux interop tests that require/procor/etc/os-release. CI on Linux remains the release gate.PYTHONPATHoverride to this next-based SDK source: 128 calendar/source unit tests and 52 reporting lifecycle integration tests passed.Activation
ESA still pins SDK
10.0.0b19; it must move to a released SDK containing this fix before activation. No ESA dependency change is included here.