Skip to content

release: v0.5.36 - #210

Merged
surlymochan merged 1 commit into
mainfrom
release/v0.5.36
Nov 16, 2026
Merged

surlymochan merged 1 commit into
mainfrom
release/v0.5.36

Conversation

@surlymochan

@surlymochan surlymochan commented Nov 16, 2026 •

Copy link
Copy Markdown
Owner

v0.5.36 — Two merged PRs, no milestone acceptance

Patch release. Two merged PRs since 0.5.35, verified end to end in CI and locally — M16 is not accepted yet, so this release carries no milestone claim.

Diagnostic sentences are not test-run claims (PR #207, merged 2026-11-16T08:40:26Z, squash b504fd4; fixes issue #201). The live-claims extractor now drops sentences reporting a command failing to run (failed spawn / ENOENT / command not found) from the claim set — they are diagnostic context, not assertions the agent made. Evidence: node test/live-claims.test.js 36 pass, 0 fail (7 new diagnostic-context tests); merged-main CI run 43833145541 success. Recovery note: the dispatch run was killed mid-flight by a VM runtime restart; the executor's uncommitted work survived in the run worktree and was verified, committed, and pushed by the operator — ledger gap disclosed in swrm #244.

pi_shim.sh executable so the openswarm arm can run (PR #209, merged 2026-11-16T14:18:45Z, squash 28d6416; fixes issue #208). bench/swebench/harness/pi_shim.sh mode 100644 → 100755, content unchanged, plus a new regression test asserting the exec bit, the committed mode, the shebang, and the OPENSWARM_PI_BIN wiring. The shim was non-executable, so every openswarm cell in the M16 MiniMax pilot died in ~1s with status unroutable ("no working pi binary") — 0/6, an infrastructure failure, not a model signal. Evidence, independently re-run by the main round on PR head 84b4a88: new test 4/4 pass; related suites + live-claims 43/43 pass; negative control (mode reverted) fails with EACCES; merged-main CI run 43884400880 success.

Honest note, stated plainly: milestone M16 is not accepted in this release — the pilot ran 24/24 complete (baseline 1/6, superpowers 4/6, openflow 1/6, openswarm 0/6), and the openswarm 0/6 was the infrastructure failure fixed by PR #209; the openswarm re-run is pending. No bench numbers are claimed in this release.

Acceptance ledger, stated plainly: the issue #5 ledger is settled as a conclusion on the MiniMax full-matrix recheck (135/135 complete, 2026-11-10) — P2-1 passed, P1-2 / P1-3 / P2-3 did not pass — so no bench numbers are claimed in this release.

Local check on this branch: full suite across all 73 test files — 817 tests, 812 pass, 4 fail, 1 skipped. The 4 failures are the same pre-existing I6d environment failures as the 0.5.35 baseline (pytest not on PATH in this VM: 1 in test/i6d-acceptance-demo.test.js, 3 in the I6d assertion-execution gate suite of test/xflow-audit.test.js, all spawnSync pytest ENOENT); the +11 tests / +11 pass over the 0.5.35 baseline (806 / 801) are exactly the 7 new live-claims tests from #207 plus the 4 new shim tests from #209 — zero regression. Note: npm test (scripts/run-tests.mjs) is fail-fast and stops at the first failing file, so the full count was measured with a single node --test test/*.test.js run; this branch changes version/docs files only. Files changed: CHANGELOG.md, RELEASE_NOTES.md, docs/launch-dossier.md, package.json, package-lock.json (same 5 files as release PR #200).

@surlymochan
surlymochan merged commit f328a59 into main Nov 16, 2026
2 checks passed
@surlymochan
surlymochan deleted the release/v0.5.36 branch November 16, 2026 14:42
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