fix(worker): wnwrap alarm DATE+TIME are UTC; stop re-applying GMTOFFSET (#226)
ci / nightly-windev (push) Has been skipped
ci / windows-x86 (push) Successful in 1m26s
ci / java (push) Successful in 2m28s
ci / portable (push) Failing after 3m38s

QueryActiveAlarms / alarm transitions reported every wnwrap alarm late by
the record's GMTOFFSET-DSTADJUST: ParseTransitionTimestampUtc treated the
XML DATE+TIME as producer-local and added the offset, but the fields are
already UTC. DEV evidence: RuntimeAlert raise EventTimeUtc 2026-07-28
20:02:46.830 (EventTimeUTCOffsetMins -300) surfaced as 2026-07-29
01:02:46.830Z. The 2026-05-01 probe capture (TIME 13:26:14.709,
GMTOFFSET 240) was committed at 09:44 EDT, so its TIME is UTC too.

GMTOFFSET/DSTADJUST are no longer read. Tests updated + #226 regressions.
This commit is contained in:
Joseph Doherty
2026-11-12 07:20:44 -05:00
parent 2b13564e04
commit 892bdc1a19
6 changed files with 91 additions and 56 deletions
+6 -1
View File
@@ -677,7 +677,12 @@ alarm-consumer surface unblocks A.2 fully. Outline:
`ACK_RTN` → in_alarm=false, acked=true).
- `DATE + TIME + GMTOFFSET + DSTADJUST` → reassemble UTC
timestamp; matches the worker's existing `Timestamp`
wire format.
wire format. *Corrected 2026-11-12 (#226): `DATE` + `TIME`
are already UTC — the 13:26:14.709 capture above was taken
at 09:26 EDT — and `GMTOFFSET`/`DSTADJUST` only describe the
producer's zone. Adding the offset made every alarm
timestamp late by it (+5 h on DEV for a `GMTOFFSET=300`
record).*
- `PRIORITY` → severity (already 1-1000-ish range).
- `TAGNAME` → reference; `PROVIDER_NAME` + `GROUP` for
scope metadata.