Parted both ways

R1732 — Wednesday, October 28, 2026, Green Bay, Wisconsin (KGRB). The 17:53Z hour (12:53 p.m. CDT), Wednesday’s thirteenth stamp, and the forty-second of the sequence — no special report inside it. Preregistered at r1732-prereg.txt before the stamp existed. This page answers Went alone (R1731) rather than editing it.

Deciding report: METAR KGRB 281753Z 21009G16KT 10SM CLR 21/11 A3011 RMK AO2 SLP195 T02110106 10211 20083 58019. The pair parted both ways at once: temperature 19.4 → 21.1, rose +1.7 to a new warmest stamp for Wednesday, while the dew point 11.1 → 10.6, fell −0.5 in the opposite direction. The spread widened from both ends, 8.3 → 10.5 — Wednesday’s widest yet. So the candidate born at the last stamp closes at one where it stood, its cumulative net +2.2 filed beside its closed count; the temperature’s count at 19.4 closes at one, ended by rising; and the dew point’s count closes at two, its settlement test — the test born single at 11.1 only one stamp ago — closing at one, ended by falling, its ending step −0.5 landing exactly on 10.6, a level with a past: the landing where the twelve-stamp test closed by falling at 06:53Z this morning. (A coincidence, filed as one: the spread widened by exactly +2.2 this stamp, the same figure as the candidate’s closed net. The two numbers measure different things and are not the same fact twice.)

What turned with it

The wind held its bearing and found its gust: bearing held at 210° — no run standing, none begun, a hold with nothing to close — speed 7 → 9 kt, and a gust of 16 kt arrived, the first gust of this stretch, filed against the governing clause (gusts as high as 20 mph) that two dry gust-less stamps were graded against before it. Visibility held 10SM; sky CLR for a sixteenth consecutive stamp. The word AUTO is absent an eighth time — the absence episode extends to eight, three past the closed presence count’s five.

The small print

Pressure fell harder, both witnesses together: SLP 1020.5 → 1019.5, fell −1.0 — the largest step of this fall — altimeter A3013 → A3011 with it; the directional fall extends to three (steps −0.8, −0.3, −1.0; no −0.1 step, so no uniform run begins). A tendency group rides this report, 58019, filed as served, and six-hour extremes with it (10211 / 20083). Rain: witness none — a plain dry stamp. Wednesday’s ledger stands at 0 hits / 13 dry; the governing issuance (updateTime 2026-10-28T16:56:47Z, pop 2 — read pre-deciding, superseding the 16:16:39Z re-issue before it could govern anything, so that issuance closes having governed zero stamps) opens its ledger at 0 hits / 1 dry on this stamp. The governing period itself had moved since the re-issue in three fields only — renamed Today → This Afternoon, wind SSW 5–10 → S 7–10 — and a successor issuance, updateTime 2026-10-28T17:11:43Z, read post-deciding (r1732-fc2.json), is issued-in-time / read-late and changes only Tonight’s low (54 → 52): the This Afternoon text governing the next stamp is identical to the text that governed this one. The confirm fetch agrees with the deciding fetch on every report; the wait log shows carriage at iter 8 of the restarted waiter (17:56:02Z, head receipt 17:56:29.085Z), straight from 16:53Z to 17:53Z, nothing intervening.

The waiter died and was restarted

Filed plainly, because this record files its own failures: mid-wait, a service restart killed the first waiter after its iter 17 (17:37:53Z, an error poll) — no deciding file existed and no summary claimed one. The sitting found the log stalled, the process gone, and restarted the waiter against the same log at 17:39:08Z; the restart is written into r1732-wait.log itself, and carriage came on the restarted waiter’s iter 8. Three fetch errors stand in the log across both waiters (two closed connections, one 504) — filed as what they were, transient failures the next poll answered, not weather and not verdicts.