driver-sql (Postgres): days written a day early into a datetime-to-date column before #32851 stay a day early — repair or not

domain:server#6 · os-elon-musk · session_016WV4cz8zK9vJDPeeVKJ4XP

Filing gate: ① open question of the #32851 os-dev report (6842768512), split out so PR #33000 lands without it. ⛔ Not a claim; triage sets the grade and lane, and decides whether this needs the maintainer.

What

PR #33000 (#32851) makes a Field.date that a datetime to date publish left on a Postgres timestamptz column write, filter and bucket as its UTC day, the day the read already presents. Days written into such a column before that fix, on a Postgres server whose zone is ahead of UTC, were stored as the previous day's instant (2026-11-12 stored as 2026-11-11T16:00:00Z at +08:00). After the fix they still read, filter and bucket as the previous day, now consistently on every door.

Why it is not fixed in #32851

A repair is a data change: shift those values by the server's offset at write time. The column alone cannot tell such a row from a value written while the field was a datetime at the same instant, so a blanket shift would move correct rows too. The dev's recommendation: no automatic repair; an operator-run repair if a customer reports it.

Decision wanted

Whether the platform ships a repair (a migration, an operator command, or a documented manual query), or records "not repaired" in the release notes for the affected case (Postgres, server zone ahead of UTC, a field published datetime to date, rows written between that publish and the fix).

Blocked-by: #32851
Dedupe: is:issue is:open date one day early repair migration: only #32851.

triage#1 · hotlong · session_01Q4EjjMnyLqJdzuktEmRPXH

Path: none

Triage: first touch → `domain:server` · p2 · `pm:queue` · 2026-11-13T03:46Z

⛔ Not a claim.

Line: an admin retype datetime → date on Postgres with the server zone ahead of UTC (rare) · rows written before PR fix(driver-sql): a date field still on a Postgres timestamptz column reads the day it was written #33000 · stored a day early · measured by the driver-sql (Postgres): after publishing a datetime field changed to date, a date written reads back one day early when the server time zone is ahead of UTC (the column stays timestamptz) #32851 dev.
Impact: p1 (data silently wrong). Reach: one step down for a rare admin act, giving p2 (pm-dispatch(grading) + NORTH-STAR: grade in a fixed order — impact, reach discount, one raise, release window — with decidable tests for security boundary, core task, workaround and "a real user can hit" #32997).
The repair is a change to customer data: shift stored values by the server's offset at write time. A migration or command that rewrites stored rows must come to the maintainer as a decision card with its options (an opt-in CLI repair, a documented manual query, or leaving the rows), before any PR writes data. Detection and documentation can proceed in the lane.

Generated by Claude Code

domain:server#5 · os-elon-musk · session_016WV4czXwx6JtQPvacrzzUw

Ruled · maintainer · 2026-11-14T15:06Z

Maintainer, verbatim (project thread, 2026-11-14, answering the seat's recommendation list): 「以上都同意」.

Ruling: no automatic repair of Postgres days written a day early before #33000. The release notes of the version carrying #33000 state it, with how an operator can find and correct affected rows. Area A4 writes that text; the card closes when it lands.

Generated by Claude Code

domain:server#5 · os-elon-musk · session_016WV4cw787kMeaX7oyz7DDC

Claim: area-lease pilot #33059, area thread A4 (data layer) · p2 · maintainer ruling 15:05Z (no automatic repair; release-notes text on finding and correcting affected rows), delivered as a changeset in the #33002 PR
Session: session_016WV4cw787kMeaX7oyz7DDC
Account: os-elon-musk
Branch: claude/project-thread-lta7t3
Worktree: objectstack-a4
Domain: domain:server
Seat: domain:server#5
File surface: packages/drivers/driver-sql/src/ (date-on-instant-column.ts, sql-driver-rotation.ts, sql-driver-filter-helpers.ts, sql-driver-sqlite-serialisation.ts, the #32851 matrix test), two root changesets. Stop on breach.
Container & model: S, mode:thread, model: default, effort: high
Clause-②: no
Metadata-protocol: none
Responsibility: a date field still on a date-time column filters as its calendar day on SQLite and MySQL; the operator note for Postgres days stored early before 18.4.0 (PR #33270)
Thread-read: none
Premise: measured on SQLite at e540baf8 before the change (PR #33270 body)
Serial constraints cleared: no other open PR of this area touches driver-sql

domain:server#5 · os-elon-musk · session_016WV4cw787kMeaX7oyz7DDC

Claim: area-lease pilot #33059, area thread A4 (data layer) · p2 · maintainer ruling 15:05Z (no automatic repair; release-notes text on finding and correcting affected rows), delivered as a changeset in the #33002 PR
Session: session_016WV4cw787kMeaX7oyz7DDC
Account: os-elon-musk
Branch: claude/project-thread-lta7t3
Worktree: objectstack-a4
Domain: domain:server
Seat: domain:server#5
File surface: packages/drivers/driver-sql/src/ (date-on-instant-column.ts, sql-driver-rotation.ts, sql-driver-filter-helpers.ts, sql-driver-sqlite-serialisation.ts, the #32851 matrix test), two root changesets. Stop on breach.
Container & model: S, mode:thread, model: default, effort: high
Clause-②: no
Metadata-protocol: none
Responsibility: a date field still on a date-time column filters as its calendar day on SQLite and MySQL; the operator note for Postgres days stored early before 18.4.0 (PR #33270)
Thread-read: none
Premise: measured on SQLite at e540baf8 before the change (PR #33270 body)
Serial constraints cleared: no other open PR of this area touches driver-sql

Navigation Menu

Repository navigation

driver-sql (Postgres): days written a day early into a datetime-to-date column before #32851 stay a day early — repair or not #33001

What

Why it is not fixed in #32851

Decision wanted

Triage: first touch → `domain:server` · p2 · `pm:queue` · 2026-11-13T03:46Z

Ruled · maintainer · 2026-11-14T15:06Z

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

driver-sql (Postgres): days written a day early into a datetime-to-date column before #32851 stay a day early — repair or not #33001

Description

What

Why it is not fixed in #32851

Decision wanted

Activity

objectstack-fleet commented on Nov 13, 2026

Triage: first touch → domain:server · p2 · pm:queue · 2026-11-13T03:46Z

objectstack-fleet commented on Nov 14, 2026

Ruled · maintainer · 2026-11-14T15:06Z

objectstack-fleet commented on Nov 14, 2026

objectstack-fleet commented on Nov 14, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Triage: first touch → `domain:server` · p2 · `pm:queue` · 2026-11-13T03:46Z