Skip to content

spec(t27os): disclosure_scan.t27 -- FOOTER_STEP_YAML, the lab step that foots every published summary (Refs #11944) - #12139

Merged
gHashTag merged 1 commit into
masterfrom
night/20-footer-step-l1
Oct 24, 2026
Merged

gHashTag merged 1 commit into
masterfrom
night/20-footer-step-l1

Conversation

@gHashTag

Copy link
Copy Markdown
Collaborator

Refs #11944 (board Trinity-S-AI-DNA/t27os#20, criterion 3)

Replaces #12136: same tree, re-committed with a local Refs #11944 because the old commit cited only the cross-repo board and failed L1. Commit f645c67; spec bytes identical to bba9dab, so lab receipt below covers it.

What

specs/t27os/disclosure_scan.t27 gets FOOTER_STEP_YAML, the lab.yml step text that appends the five disclosure lines (reader, judge, n, cost, repro) to the published runs/<sha>.<c8>.summary.txt.

Why

Measured 2026-10-25: none of the 110 runs/*.summary.txt files in Trinity-S-AI-DNA/t27os has a footer. The checker in this spec would report 110 gaps on them. The writer is what closes those gaps.

Pinned by tests

  • All five keys are present, in footer order.
  • Only the published copy gets the footer. /tmp/out/summary.txt stays as rows, because the typed-result step and the fail gate read it.
  • cost is measured from run_started_at. If the start time cannot be read, the run fails instead of publishing a made-up cost.
  • The step's output for receipt fd0ee966b3 (lab run 40136220581, 278 s) gets COV_ALL. Negative control: the same receipt as it is published today gets COV_GAP.

Verified

Lab receipt runs/bba9dabd872b12b3e5fb67664e5401233cef3107.0bee3808.json (run 40142947612), result.ok=true:

  • t27c: disclosure_scan 13/13, disclosure 6/6, 0 vacuous, 0 blocked
  • t27b: disclosure_scan 13/13 (asserts 52), disclosure 6/6 (asserts 21)

Next: the copy half goes into t27os lab.yml, proven by a branch-ref lab run.

🤖 Generated with Claude Code

…p that foots every published summary (Refs #11944)

Same tree as bba9dab (PR #12136), re-committed because that commit cited
only the cross-repo board Trinity-S-AI-DNA/t27os#20 and failed L1.
Lab receipt runs/bba9dabd872b12b3e5fb67664e5401233cef3107.0bee3808.json
(run 40142947612): t27c 13/13 + 6/6, t27b 13/13 + 6/6, ok=true.

Refs #11944

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-24 23:41:20 UTC

Summary

Status Count
Total Open PRs 44
PRs with Failing Checks 40
PRs with All Checks Green 4
READY 2
FAILING 40
PENDING 0
NO CHECKS YET 0

These columns do not partition: 2 + 40 + 0 + 0 = 42, and there are 44 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=55c0f1b1b32d != 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).

@tri-claw

tri-claw Bot commented Oct 24, 2026

Copy link
Copy Markdown

Summary by t27-bees

Checked by the t27 compiler

t27c parse, parse-complete and typecheck on each changed .t27 file at f645c67. Nothing was generated, built or run. A pass means the file is well formed, not that it is right.

  • specs/t27os/disclosure_scan.t27: parses and typechecks.

Read by the model

This PR adds FOOTER_STEP_YAML, a GitHub Actions workflow step that appends five disclosure footer lines (reader, judge, n, cost, repro) to published summary files in runs/. It includes the step YAML as a string constant, example rendered output, and tests verifying the footer content, ordering, target file selection, cost measurement with failure on missing start time, and coverage verdicts (footed = COV_ALL, unfooted = COV_GAP).

Worth a look

  • specs/t27os/disclosure_scan.t27:48–78 (FOOTER_STEP_YAML) – The gh api call depends on gh CLI being present and GH_TOKEN having actions:read scope; on self-hosted runners without gh, the step will fail silently (exit 1 with error log). The date -u -d "$t0" uses GNU date -d flag; this works on Ubuntu runners but would break on macOS runners.
  • specs/t27os/disclosure_scan.t27:50 – sha=${{ steps.sha.outputs.resolved }} assumes a prior step named sha with output resolved; if that step is missing or renamed, sha expands to empty string and the git ls-files pattern becomes runs/.*.summary.txt, potentially matching the wrong file.
  • specs/t27os/disclosure_scan.t27:51 – git ls-files -o --exclude-standard only finds untracked files; if the summary file is already committed or not in the working tree (e.g., sparse checkout), sum will be empty and the step exits with "no summary to foot".
  • specs/t27os/disclosure_scan.t27:55 – /tmp/t27c_version.txt is read for the reader line; if the prior step that writes it fails or runs on a different runner (matrix), the file won't exist and head -1 outputs nothing, producing a malformed reader line.
  • specs/t27os/disclosure_scan.t27:59 – The repro command contains -f challenge=<fresh hex> as a literal placeholder; publishing this verbatim gives users a non-runnable command unless they manually substitute a real challenge value.

Read: 1 files, +55 -0, the whole diff. Head f645c67. Model: openai-compatible/nvidia/nemotron-3-ultra-550b-a55b. Commands: @t27-bees review | summary | help | pause | resume. Free for every repository.

@gHashTag
gHashTag merged commit 93948c4 into master Oct 24, 2026
37 checks passed
gHashTag added a commit to Trinity-S-AI-DNA/t27os that referenced this pull request Oct 25, 2026
)

Places FOOTER_STEP_YAML from Trinity-S-AI-DNA/t27#12139 (specs/t27os/disclosure_scan.t27)
between the typed result step and publish, rendered from the spec string.
The published runs/<sha>.<c8>.summary.txt gains reader/judge/n/cost/repro;
/tmp/out/summary.txt, which the gates read, stays rows only.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
@gHashTag
gHashTag deleted the night/20-footer-step-l1 branch October 26, 2026 03:43
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