Skip to content

spec(crons): the lab.yml encoder step as t27 law -- M6 only, distro ffmpeg first, never fatal (Refs #11873) - #11891

Merged
gHashTag merged 6 commits into
masterfrom
night/10-encoder-step
Oct 26, 2026
Merged

gHashTag merged 6 commits into
masterfrom
night/10-encoder-step

Conversation

@gHashTag

Copy link
Copy Markdown
Collaborator

Refs #11873 -- see Trinity-S-AI-DNA/t27os#10 (lab.yml loses receipts on a third-party ffmpeg CDN).

specs/crons/t27os-lab-loop.t27 is the spec whose SERVICE is .github/workflows/lab.yml. Hand-written workflow YAML is refused by the language rule, so this PR puts the source of the "check the encoder plumbing" step here, the same pattern as specs/tools/t27o/install.t27 WORKFLOW_YAML: TRI CLAW copies ENCODER_STEP_YAML verbatim over the step.

The step's law, checked by encoder_step_ok:

  1. M6 only -- nothing is fetched unless the dispatch asked for the encode (m6 = "true"); today the step fetches on every dispatch, so an m6=false run still dies on the CDN.
  2. Distro ffmpeg first, then apt, and only then the static build (the distro 6.1.1 encoded the first receipt of 2026-10-23).
  3. The static build is downloaded to a file with --retry 3, never piped into tar (run 39980640303: xz: (stdin): File format not recognized), and a failed fetch is written to /tmp/ffmpeg.txt, never fatal.

Tests: the new step keeps all three; negative control -- today's step (t27os master 234d76e, lab.yml lines 99-111, kept as ENCODER_STEP_YAML_OLD) is refused.

Complements #11884 (the surface_runs / receipt_written decisions in specs/t27os/labs.t27); this is the concrete step text, disjoint file.

Verification: t27c typecheck OK; lab receipt (t27c leg) will be posted below. Draft until the t27b leg exists (t27#11889 / #11890).

🤖 Generated with Claude Code

…fmpeg first, static fetch to a file and never fatal (Refs #11873)

See Trinity-S-AI-DNA/t27os#10: the step piped the johnvansickle static
download into tar on every dispatch and a CDN hiccup killed the run
before the receipt. ENCODER_STEP_YAML is what TRI CLAW copies over the
step; encoder_step_ok checks the three rules; today's step is the
negative control.

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

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-23 20:56:38 UTC

Summary

Status Count
Total Open PRs 38
PRs with Failing Checks 34
PRs with All Checks Green 4
READY 4
FAILING 34
PENDING 0
NO CHECKS YET 0

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=5aec047d42ec != 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

Copy link
Copy Markdown
Collaborator Author

Lab receipt (t27c leg) for 0d353fd: Trinity-S-AI-DNA/t27os runs/0d353fdde73fb49fed65ea8dba5c8a5448807343.cef4b002.json (run 40022937699, success)
pass/fail: specs/crons/t27os-lab-loop.t27 tests=4 pass=4 FAIL=0 vacuous=0 blocked=0.
The receipt itself shows the bug: this was an m6=false dispatch and its ffmpeg field is still 7.0.2-static https://johnvansickle.com/... -- today's step fetched the CDN build anyway. t27b leg still missing (t27#11889); PR stays draft.

@gHashTag

Copy link
Copy Markdown
Collaborator Author

Night tick 27, t27b leg for head 0d353fd (t27b-native-ratchet run 40022996412, artifact t27b-native-run/run.json, ubuntu-24.04-arm):

  • specs/crons/t27os-lab-loop.t27: t27b pass, 4 tests / 10 asserts (reference pass)

The check is red for ledger staleness only: 0 UNEXPECTED FAILURE, 23 UNEXPECTED PASS, 54 UNLISTED. Master f308c93 is red the same way (run 40001627259), so this red comes from master, not from this PR. Both compiler legs are now on record (t27c from the T27 OS lab receipt, t27b here). Refs #11889.

This was referenced Oct 23, 2026
This was referenced Oct 25, 2026
@gHashTag
gHashTag marked this pull request as ready for review October 26, 2026 01:46
@gHashTag
gHashTag merged commit 11809ca into master Oct 26, 2026
33 of 37 checks passed
@gHashTag
gHashTag deleted the night/10-encoder-step branch October 26, 2026 03:45
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