Skip to content

fix(tri): a check that never ran is not a check that never failed (Closes #2276) - #2277

Merged
gHashTag merged 1 commit into
masterfrom
fix/prcheck-no-baseline-v2
Aug 20, 2026
Merged

fix(tri): a check that never ran is not a check that never failed (Closes #2276)#2277
gHashTag merged 1 commit into
masterfrom
fix/prcheck-no-baseline-v2

Conversation

@gHashTag

Copy link
Copy Markdown
Owner

Closes #2276.

tri pr ready built its baseline from failures elsewhere. A check absent from that baseline was printed as NOT failing on recent master commits or in the last 5 merged PRs — which reads the same whether the check ran there and passed, or never ran there at all.

emit-bitexact-gate.yml is the second kind: pull_request with a paths: filter and no push: trigger. 0 master runs ever; 0 of the last 5 merged pull requests ran it. The verdict came out of zero observations in both windows and was printed as evidence.

Now the command records what was observed, not only what was red, and the third state gets its own verdict:

  emit-bitexact
      NO BASELINE — this check did not run on any recent
      master commit nor on any of the last 5 merged PRs, so
      there is nothing to compare against. Usually a `paths:` filter
      with no `push:` trigger. Read the log; this command cannot say
      whether the failure is yours.

VERDICT: CANNOT TELL — 1 failure(s) have no baseline to compare against:

It still refuses to merge — but it now says the failure is a finding about the repository's CI, not about the change. A regression test asserts the two states classify differently.

🤖 Generated with Claude Code

…oses #2276)

tri pr ready classified failures against a baseline of FAILURES elsewhere.
A check missing from that baseline was printed as 'NOT failing on recent
master commits' — a sentence that reads identically whether the check ran
there and passed, or never ran there at all.

emit-bitexact-gate.yml is the second kind: pull_request with a paths:
filter and no push: trigger. Zero master runs, ever; zero of the last five
merged pull requests touched the filtered paths. The verdict was assembled
from zero observations in both windows and printed as evidence — the same
shape as every other instrument in this project that reported what it could
not see as a clean reading.

Records what was OBSERVED, not only what was red, and gives the third state
its own verdict: NO BASELINE / CANNOT TELL. It still blocks the merge, but
names the cause as the repository's CI rather than the change under review.
@github-actions

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-08-20 01:08:58 UTC

Summary

Status Count
Total Open PRs 21
PRs with Failing Checks 6
PRs with All Checks Green 15
READY 7
FAILING 6
PENDING 0

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=cbbfac87dff3 != 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).

@github-actions

Copy link
Copy Markdown
Contributor

📓 NotebookLM Notebook linked to this PR

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

@gHashTag
gHashTag merged commit 39f93d1 into master Aug 20, 2026
29 of 34 checks passed
@gHashTag
gHashTag deleted the fix/prcheck-no-baseline-v2 branch August 20, 2026 01:25
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 pr ready read 'never ran' as 'never failed' — a verdict built from zero observations

1 participant