4856638097
A day later, the same functions:log query showed generation=2 and =3 HAD fired
at the exact times of the "failed" rotations. The stranding was real (the
partner sat on 🔒 across cold restarts), but the client-side mechanism I wrote
down as measured was inferred from log silence — and the CLI's log window is
paginated and lagging, so silence proves nothing. The entry now says what was
observed, what the logs later showed, what remains unpinned, and why the fixes
(server-authoritative read; rules interlock) are correct hardening regardless.
New standing lesson recorded: never conclude "X didn't fire" from functions:log
absence; prove absence with a side effect.
|
||
|---|---|---|
| .. | ||
| brand | ||
| crypto | ||
| qa | ||
| release | ||
| screenshots | ||
| store | ||
| CopyMigration.md | ||
| Engineering_Reference_Manual.md | ||
| NameChange.md | ||
| copy-guide.md | ||
| date-planning-roadmap.md | ||
| standardization-review.md | ||