Skip to content

feat(tri): one-away, the specs a single repair can actually move - #3360

Merged
gHashTag merged 4 commits into
masterfrom
corpus-one-away
Sep 6, 2026
Merged

feat(tri): one-away, the specs a single repair can actually move#3360
gHashTag merged 4 commits into
masterfrom
corpus-one-away

Conversation

@gHashTag

@gHashTag gHashTag commented Sep 6, 2026

Copy link
Copy Markdown
Owner

Closes #3359

specs read                    650
rustc accepted                352
did not GENERATE              69   (not zero errors -- a different fact)
rustc refused                 229

real errors per failing spec, the summary line excluded:
    1: 50   2: 34   3: 18   ...

50 spec(s) are REPORTED as carrying exactly 1 error.
   24  the count is exact -- every diagnostic carries an [E####] code,
       so rustc ran its passes to completion.
   26  the count is a LOWER BOUND -- the first error has no code, which
       means rustc stopped parsing there. Repair it and the next appears.

Two corrections are baked in, and each was found by being wrong first.

The summary line. rustc ends with error: aborting due to N previous errors, which matches ^error like any diagnostic. Counting it inflates every bucket by one and turns specs with one error into zero specs with one error — the opposite conclusion, and the one written down first in #3349.

The lower bounds. The command first reported 56 one-error specs. Taking one — server/http.t27, sole error expected expression, found keyword \fn`— and repairing it revealedexpected expression, found `@`underneath. rustc abandons a file at the first **parse** error, so a sole diagnostic with no[E####]` code is a floor, not a total. The honest headline is 24, not 50. The rule needs no list of message shapes: the presence of an error code is the answer to "did the compiler keep going".

Eight controls, one per claim — including one asserting that counting the summary would give the opposite answer, and one for each side of the code/no-code split, so neither mistake can come back silently. Six unit tests over the three pure functions. A spec that does not GENERATE is reported separately; it is not a spec with zero errors.

…#3359)

The measurement that explained three +0 iterations lived in a shell script in
/tmp. A command survives the session; a script does not.

Census re-blessed in this same commit, as the gate demands: `fetches` moved
`files read` 44 -> 45, which is this commit's own new file
`cli/tri/src/oneaway.rs` and nothing else.
@github-actions

github-actions Bot commented Sep 6, 2026

Copy link
Copy Markdown
Contributor

📓 NotebookLM Notebook linked to this PR

This notebook contains session context, decisions, and artifacts for this work.

@github-actions

github-actions Bot commented Sep 6, 2026

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-09-06 05:17:13 UTC

Summary

Status Count
Total Open PRs 16
PRs with Failing Checks 10
PRs with All Checks Green 6
READY 3
FAILING 10
PENDING 0
NO CHECKS YET 0

These columns do not partition: 3 + 10 + 0 + 0 = 13, and there are 16 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=23f03e8a97d5 != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

Census re-blessed in this same commit, as the gate demands; the numbers moved
with the commits being merged in, not with this branch's own work.
@github-actions

github-actions Bot commented Sep 6, 2026

Copy link
Copy Markdown
Contributor

📓 NotebookLM Notebook linked to this PR

This notebook contains session context, decisions, and artifacts for this work.

@github-actions

github-actions Bot commented Sep 6, 2026

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-09-06 05:20:14 UTC

Summary

Status Count
Total Open PRs 16
PRs with Failing Checks 10
PRs with All Checks Green 6
READY 4
FAILING 10
PENDING 0
NO CHECKS YET 0

These columns do not partition: 4 + 10 + 0 + 0 = 14, and there are 16 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=23f03e8a97d5 != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

)

rustc abandons a file at the first parse error, so for a spec whose sole
diagnostic carries no [E####] code the count is a floor, not a total. Of 50
reported as one-error on today's master, 24 are exact and 26 are lower bounds.
Found by repairing one of them and watching the next error appear.
@github-actions

github-actions Bot commented Sep 6, 2026

Copy link
Copy Markdown
Contributor

📓 NotebookLM Notebook linked to this PR

This notebook contains session context, decisions, and artifacts for this work.

@github-actions

github-actions Bot commented Sep 6, 2026

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-09-06 06:14:40 UTC

Summary

Status Count
Total Open PRs 14
PRs with Failing Checks 10
PRs with All Checks Green 4
READY 3
FAILING 10
PENDING 0
NO CHECKS YET 0

These columns do not partition: 3 + 10 + 0 + 0 = 13, and there are 14 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=23f03e8a97d5 != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

Census re-blessed in this same commit, as the gate demands; the numbers moved
with the commits being merged in, not with this branch's own work.
@github-actions

github-actions Bot commented Sep 6, 2026

Copy link
Copy Markdown
Contributor

📓 NotebookLM Notebook linked to this PR

This notebook contains session context, decisions, and artifacts for this work.

@github-actions

github-actions Bot commented Sep 6, 2026

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-09-06 06:17:03 UTC

Summary

Status Count
Total Open PRs 14
PRs with Failing Checks 10
PRs with All Checks Green 4
READY 3
FAILING 10
PENDING 0
NO CHECKS YET 0

These columns do not partition: 3 + 10 + 0 + 0 = 13, and there are 14 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=23f03e8a97d5 != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

@gHashTag
gHashTag merged commit aeb8cdb into master Sep 6, 2026
30 checks passed
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.

tri one-away: the specs a single repair can actually move

1 participant