Skip to content

One fact, one number: replace the forked deploy run count with a dated receipt - #8730

Merged
briansrls merged 1 commit into
mainfrom
session/nimble-fox-671-countfix
Aug 21, 2026
Merged

briansrls merged 1 commit into
mainfrom
session/nimble-fox-671-countfix

Conversation

@gunbai-bot

@gunbai-bot gunbai-bot Bot commented Aug 21, 2026

Copy link
Copy Markdown
Contributor

Two annotations described the same deleted deploy workflow with different numbers — fleet_desired_admission_workflow said 14 runs, fleet_desired_admission_witness_test said 30. One fact with two numbers is a §3 fork, and it landed in #8643 with both halves intact.

Measured rather than picking a side, from the run history of the deleted workflow:

30 runs, 2026-08-20T03:55:43Z .. 2026-08-20T17:14:32Z
29 failure, 1 cancelled, 0 success

So 14 was wrong — and exited 127 on all 30 was also wrong, because one run was cancelled and never reached the shell at all. Both annotations now carry the dated window instead of a bare count.

Why a receipt rather than a corrected number. A count measured once and then restated as a timeless fact decays silently: nothing re-derives it, nothing contradicts it when it drifts, and the next person to measure gets a third number — which is exactly how this fork was produced. A dated window is falsifiable against a stated observation. The subject is a deleted workflow, so the window is closed and the receipt cannot go stale.

Prose-only; no semantic change, no emitted-artifact change. Annotations stay at module-item grain (§4c) — verified no indented // was introduced.

— sent from nimble-fox-671

…d receipt

Two annotations described the same deleted deploy workflow with different
numbers -- the workflow annotation said 14 runs, the witness annotation said 30.
One fact with two numbers is a §3 fork, and a count restated as a timeless fact
decays the moment anything changes.

Measured from the run history rather than picking a side: 30 runs between
2026-08-20T03:55:43Z and 2026-08-20T17:14:32Z, 29 failure and 1 cancelled, zero
successes ever. So 14 was wrong -- and "exited 127 on all 30" was also wrong,
because one run was cancelled and never reached the shell at all.

Both annotations now carry the dated window, so the claim is falsifiable against
a stated observation instead of asserting a number nobody can re-derive.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@briansrls
briansrls merged commit 2249df7 into main Aug 21, 2026
1 check failed
@briansrls
briansrls deleted the session/nimble-fox-671-countfix branch August 21, 2026 05:26
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant