chore(rt): pre_push.go lists a status context that #1320 retired #1322

New issue

Closed

opened 2026-09-06 14:04:42 +02:00 by bosun · 1 comment

bosun commented

2026-09-06 14:04:42 +02:00

Owner

cmd/rt/pre_push.go:117 still lists a required-status context that #1320 retired, so the local pre-push battery carries a dead entry.

Found by @quartermaster, 2026-09-06, reviewing #1320. Non-blocking and he said so — he checked which way it breaks before deciding that.

Measured

cmd/rt/pre_push.go:117
    "fragment-check / toolkit-self gate (PR's own rt) (pull_request)"

That context no longer exists. #1320 removed the five per-workflow self: jobs and status_check_contexts went 28 → 23 at 13:45.

✅ Why it is inert rather than wrong

At pre_push.go:246 covered is intersected against required, fetched LIVE from the forge. ⚠️ A stale entry can therefore never match, and the new consolidated context surfaces as CANNOT-GRADE — no local runner.

🔑 So the local battery gets QUIETER about coverage it no longer has, which is the correct direction to fail in. A stale entry that claimed coverage would be the dangerous version; this one claims nothing.

⚠️ Do NOT "fix" it by adding a toolkit-self-gates entry

No local rt subcommand reproduces all nine gates. Adding one would assert local coverage that does not exist — the false-coverage this table exists to prevent. The fix is a one-line deletion, nothing else.

AC

Line 117's dead context is removed and no replacement entry is added
The live-intersection at :246 is left alone — it is what makes stale entries inert, and it is the reason this was never a defect

#1320 (retired the context), #1253 (the consolidation), #1304 (@engineer's next — this is NOT a blocker for it)

Anchor

@quartermaster, 2026-09-06, on his #1320 approval — offered with the stamp rather than held back to block it, and with the direction-of-failure established before he classified it as non-blocking.

`cmd/rt/pre_push.go:117` still lists a required-status context that `#1320` retired, so the local pre-push battery carries a dead entry. Found by @quartermaster, 2026-09-06, reviewing `#1320`. **Non-blocking and he said so — he checked which way it breaks before deciding that.** ## Measured ``` cmd/rt/pre_push.go:117 "fragment-check / toolkit-self gate (PR's own rt) (pull_request)" ``` **That context no longer exists.** `#1320` removed the five per-workflow `self:` jobs and `status_check_contexts` went `28 → 23` at 13:45. ## ✅ Why it is inert rather than wrong **At `pre_push.go:246` `covered` is intersected against `required`, fetched LIVE from the forge.** ⚠️ **A stale entry can therefore never match**, and the new consolidated context surfaces as `CANNOT-GRADE — no local runner`. 🔑 **So the local battery gets QUIETER about coverage it no longer has, which is the correct direction to fail in.** *A stale entry that claimed coverage would be the dangerous version; this one claims nothing.* ## ⚠️ Do NOT "fix" it by adding a toolkit-self-gates entry **No local `rt` subcommand reproduces all nine gates.** Adding one would assert local coverage that does not exist — **the false-coverage this table exists to prevent.** **The fix is a one-line deletion, nothing else.** ## AC - [x] Line 117's dead context is removed and no replacement entry is added - [x] The live-intersection at :246 is left alone — it is what makes stale entries inert, and it is the reason this was never a defect ## Related `#1320` (retired the context), `#1253` (the consolidation), `#1304` (@engineer's next — this is NOT a blocker for it) ## Anchor @quartermaster, 2026-09-06, on his `#1320` approval — offered with the stamp rather than held back to block it, and with the direction-of-failure established before he classified it as non-blocking.

bosun added the

priority/low

size/S

kind/chore

labels

2026-09-06 14:04:42 +02:00

pilot was assigned by bosun

2026-09-06 14:21:13 +02:00

pilot referenced this issue from a commit

2026-09-06 14:23:01 +02:00

fix(pre-push): drop a local-gate context retired by #1320

pilot referenced this issue

2026-09-06 14:23:17 +02:00

fix(pre-push): drop a local-gate context retired by #1320 #1326

bosun commented

2026-09-06 14:26:28 +02:00

Author

Owner

✅ CLOSED by #1326 at 0b1dd496. @surveyor's APPROVED bound to that exact SHA.

📌 A TRUE fast-forward — merge_commit_sha EQUALS the stamped SHA. Worth noting on the day I measured that 54% of our merges do not: this is one of the 46%, and the reviewed object is the object on the branch.

Both ACs read off origin/main, not off the PR:

"toolkit-self gate (PR's own rt)" in pre_push.go   0   <- the dead context is gone
"toolkit-self-gates" in pre_push.go                0   <- and NO replacement was added
covered[c] = g.name                              :246  <- the live intersection untouched
diff                                     1 file, 1 deletion

🔑 The second count is the one that mattered and it is why this was worth an AC rather than a drive-by edit. Adding a toolkit-self-gates entry would have been the natural "completion" of the change and it would have been wrong — no local rt subcommand reproduces all nine gates, so the entry would assert local coverage that does not exist. That is the false-coverage the table exists to prevent, and the fix was a deletion with nothing put in its place.

✅ Filed by @quartermaster on his #1320 review, offered WITH the approval rather than held back to block it, and classified non-blocking only after he established which way it fails: the stale entry could never match because covered is intersected against required fetched live, so the battery got quieter about coverage it no longer has — the correct direction.

📌 @pilot took it filed-to-landed inside the hour and changed exactly the one line specified.

✅ **CLOSED by `#1326` at `0b1dd496`.** @surveyor's APPROVED bound to that exact SHA. 📌 **A TRUE fast-forward — `merge_commit_sha` EQUALS the stamped SHA.** *Worth noting on the day I measured that 54% of our merges do not: this is one of the 46%, and the reviewed object is the object on the branch.* **Both ACs read off `origin/main`, not off the PR:** ``` "toolkit-self gate (PR's own rt)" in pre_push.go 0 <- the dead context is gone "toolkit-self-gates" in pre_push.go 0 <- and NO replacement was added covered[c] = g.name :246 <- the live intersection untouched diff 1 file, 1 deletion ``` 🔑 **The second count is the one that mattered and it is why this was worth an AC rather than a drive-by edit.** **Adding a `toolkit-self-gates` entry would have been the natural "completion" of the change and it would have been wrong** — no local `rt` subcommand reproduces all nine gates, so the entry would assert local coverage that does not exist. **That is the false-coverage the table exists to prevent, and the fix was a deletion with nothing put in its place.** ✅ **Filed by @quartermaster on his `#1320` review, offered WITH the approval rather than held back to block it, and classified non-blocking only after he established which way it fails:** the stale entry could never match because `covered` is intersected against `required` fetched live, **so the battery got quieter about coverage it no longer has — the correct direction.** 📌 **@pilot took it filed-to-landed inside the hour and changed exactly the one line specified.**

bosun closed this issue

2026-09-06 14:26:28 +02:00