Skip to content

spec(t27f): frame_render -- display list ops to PPM frames, the M6 render harness (refs #11781, #10265) - #11816

Merged
gHashTag merged 1 commit into
masterfrom
lane/t27f-frame-render-m6
Oct 23, 2026
Merged

gHashTag merged 1 commit into
masterfrom
lane/t27f-frame-render-m6

Conversation

@gHashTag

Copy link
Copy Markdown
Collaborator

The missing M6 piece between export_video.t27's records and the mp4: a pure rasterizer for the display list ops into PPM frames at the export presets (1920x1080 / 1080x1920), plus the P6 writer -- ffmpeg consumes PPM sequences directly, so the lab driver loops frame_bytes through ffmpeg with no other new plumbing.

  • 9/9 tests, 0 FAIL, 0 vacuous; 2 invariants proved; 267 runtime asserts. Sealed, all hashes MATCH. The dupe corpus went DOWN one (reuses t27os editors' pub same_bytes instead of a 23rd private copy).
  • Renders exactly: fill_rect, alpha (Q8, the recorded rule), translate groups (composed, real restore) -- and REFUSES stroke/path/text/image/clip by name with the buffer untouched (a dry validation pass runs before any paint; export_video's own fixture stream refuses E_UNSUPPORTED -- its text/face are named, not hidden; the pixel projection of the same 900-frame motion law is what frame_bytes renders, test-enforced split).
  • Determinism: frame 0 vs 450 differ; two renders of the same frame are byte-identical AND sha256-equal (the work_graph digest discipline).
  • Reuses export_video::frame_stream -- no second flattener (FR-001 of spec(t27f): export_video -- same display list to MP4 on the lab #10265).

NOT CLAIMED: stroke/text/image/clip rendering, scale (needs a display_list encoding change), PNG, the ffmpeg invocation and the MP4 -- the lab loops frame_bytes at the presets for k in 0..899; LAB_RUN_DONE stays false until that dispatch.

phi^2 + 1/phi^2 = 3 | TRINITY

@github-actions

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-23 13:31:46 UTC

Summary

Status Count
Total Open PRs 28
PRs with Failing Checks 25
PRs with All Checks Green 3
READY 3
FAILING 25
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
gHashTag merged commit bebd2b1 into master Oct 23, 2026
24 of 29 checks passed
@gHashTag
gHashTag deleted the lane/t27f-frame-render-m6 branch October 26, 2026 03:46
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