Skip to content

Publish sanitized DS4 W2 release receipts - #8

Open
OmarB97 wants to merge 4 commits into
kacper-daftcode:mainfrom
OmarB97:work/ds4-w2-4-public-receipts
Open

Publish sanitized DS4 W2 release receipts#8
OmarB97 wants to merge 4 commits into
kacper-daftcode:mainfrom
OmarB97:work/ds4-w2-4-public-receipts

Conversation

@OmarB97

@OmarB97 OmarB97 commented Jul 12, 2026

Copy link
Copy Markdown

Why

The release claims need reviewable raw receipts without publishing host topology, runtime identifiers, private paths, or protected pack/checkpoint fingerprints.

What changed

  • Publishes sanitized baseline, P0 cold-restage, P1 32K, and P2 128K receipts.
  • Separates the comprehensive 41d7b2f qualification from the final 241ba984 focused sentinel.
  • Includes cold-only memory traces, per-seed manifests/raw outputs, context receipts, cleanup state, and exact frozen-RC aggregate data.

How to review

Treat these 45 files as generated evidence. Check aggregate/manifests against their raw JSONL rows, inspect the P0 trace boundaries, and confirm that the frozen-RC sentinel does not relabel the earlier three-seed/three-depth series.

Evidence

All JSON and JSONL parse, all TSV rows retain a consistent column count, and the privacy scan finds no host alias, private address/path, opaque runtime ID, or credential pattern. The final checksum/index is intentionally added by the following narrative PR after these payloads land.

Verification

  • jq empty on every JSON file
  • jq -c . on every JSONL row
  • TSV column-consistency checks
  • Public privacy-pattern scan
  • git diff --check

Risks / gaps

This is deliberately a generated-data-only PR, so its line count is not hand-authored implementation complexity. The human narrative, cross-links, exact-image index, and 47-file checksum manifest are isolated in the final PR.

Collaborators

  • Omar Baradei — operator and publication boundary.
  • Codex — sanitization, receipt reconciliation, and PR preparation.

9prodhi added a commit to 9prodhi/vLLM-Moet that referenced this pull request Jul 13, 2026
…sized-resume truncate, chmod fsync, no monitor/fd leak

Re-review wave 2 (5 items on tools/plane_file.py, tools/build_plane_file.py,
tools/test_plane_file.py):

- [kacper-daftcode#7] CLI verify/build catch ALL exceptions at the boundary (KeyError/
  TypeError from malformed-but-checksum-valid metadata, OSError) and exit
  non-zero with a one-line clean error; tracebacks behind a new --debug flag.
- [kacper-daftcode#8] RSS budget is a HARD gate: _enforce_budget() aborts the build BEFORE
  finalization when the sampled monitor observes the budget exceeded; the .tmp
  is kept for resume and no finalized artifact is produced.
- [kacper-daftcode#10] finalize() fsyncs the file inode after os.chmod(0o444) so the mode
  change is durable, not just the rename.
- [new Important] resume normalizes an OVERSIZED .tmp: ftruncate back to the
  exact layout length (was: grow-only), never finalize a file verify rejects.
- [new Minor] mon.start() + PlaneFileWriter construction moved inside the
  try/finally, monitor thread named + stop() guarded, and the writer ctor
  closes its fd if post-open validation raises: a resume-validation failure
  leaks neither the monitor thread nor an fd.

TDD (failing test first) for kacper-daftcode#7, kacper-daftcode#8, and the oversized-resume item; regression
tests for the leak item. Full suite: 73 passed, 0 failed (golden on CUDA).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
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