[Nonary]

Summary

Port two existing master fixes to canary without unrelated master changes:

• c292542: pass an explicit empty stream environment to the supervised RTX HDR foreground helper.
• 83bb316: align the extracted route test with the seven-argument API and assert that no environment entries are supplied.

The Arch failure visible on #530 is inherited from canary, not introduced by its WGC changes. At investigation time, its broker file was byte-identical to canary e686f72. The RTX HDR port retained Vibeshine's five-argument call, but Vibepollo already had the seven-argument stream-environment API. This PR does not modify #530 or Vibeshine.

Failure evidence: https://github.com/Nonary/Vibepollo/actions/runs/39883969908/job/120187744557

Attribution and publication

These are byte-equivalent ports of the existing master fixes, credited to original author Chase Payne through explicit Co-authored-by trailers and source-commit references. They were published through the authenticated connector, so their commit identities/hashes differ from the local cherry-picks; no claim is made that the original Git author fields were retained. Both published file blobs and the complete final tree were verified against the tested local port.

Validation

• Actual C broker compile/link, with and without VIBEPOLLO_HAVE_WAYLAND_OBSERVER=1: passed.
• Full test_session_exec_policy C test: passed.
• Full test_managed_app_focus_adapters.py, including supervised foreground route, Playnite branches, Windows helper ports and read-only KWin observation: passed.
• test_machine_service_contract.py: passed.
• git diff --check: passed.
• Negative control: restoring the old call fails actual C compilation.
• Negative control: corrected route fixture rejects the old five-argument call.
• Negative control: nonzero stream-environment count trips the fixture assertion.

Validation used the real libcap runtime and Debian libcap development headers extracted into the isolated test workspace. Existing unrelated warning diagnostics remain. Full Arch packaging and installed desktop integration were not run locally; CI remains to be checked on this exact published head.

Draft for independent review. No merge, release, tag, or deployment is requested.

[Nonary]

Round 1 independent review — clear, no actionable findings in the complete two-file diff.

• Session: 01a16697-602e-755b-a6cc-5e0630a570ff
• Head: 7d23d8131daa567756e29b3424e39a7df53be7f2
• Base (canary): e686f72e336494078131784db4f43670c063f49f
• Verified execution: installed C:/Users/Chase/.codex/skills/ultracode/SKILL.md; native helper model gpt-6-luna, reasoning high, confirmed available in host controls/cache and used in successful spawn calls. 6 independent review packets completed within the requested 72 total ceiling, with at most six concurrent helpers. The installed skill is a general workflow, with no named “72” preset; this does not claim 72 completed reviews. No nontrivial bug candidate was selected, so four-pair candidate adjudication was not triggered.

The production call now matches the actual seven-argument API. 0, NULL passes its stream-environment validator, adds no stream variables, and retains the selected session environment and supervised helper path. The fixture matches the real signature and asserts both empty-environment values. Both patches and final changed blobs exactly match master source fixes c29254210cf1604d2980dd99556e4fd0baf5bd69 and 83bb316804190cd2919c6477e630bdd2b129bdac.

Validation in this session: full test_managed_app_focus_adapters.py passed using Windows MSYS2 G++ 16.2 and Node, including the extracted foreground route, Playnite branches, OS-port helpers and KWin script. Generated-harness negative controls rejected the old five-argument call at compilation and rejected nonzero count, nonnull array, and nonempty environment at runtime. These controls changed only generated test harnesses. Independently checked exact-head Arch package CI: successful, including Build PKGBUILD. The PR CI run succeeded; check runs are completed success/skipped.

Limits: this Windows review did not repeat native Linux broker C compilation, Linux policy/machine-service tests, or installed desktop integration. Prior real C compilation/behavior results are reported by the implementation and were not re-labelled as this session's execution. Hosted Linux artifact validation is skipped in CI. Local Git inspection of C:/Users/Chase/sources/Vibepollo was denied; that path was stopped, and SHA-pinned GitHub sources were used instead. Both downloaded changed files hash to the published blobs. The default sandbox temp location also denied test-file creation; testing completed under the permitted task workspace.

Head/base, base branch tip and existing PR comments were rechecked immediately before posting and remain unchanged; no duplicate review existed. This is a COMMENT review from the shared account. The PR remains draft; round 1 makes no readiness transition or merge decision and waives no enforced repository check.

AI-generated review under the authorized bugfix review workflow.

[Nonary]

Final independent review (round 2) — clear, no actionable findings in the complete current two-file diff.

• Session: 01a166a0-6b9a-70c8-a3d7-11fae982b0c9
• Head: 7d23d8131daa567756e29b3424e39a7df53be7f2
• Base (canary): e686f72e336494078131784db4f43670c063f49f
• Installed UltraCode workflow read at C:/Users/Chase/.codex/skills/ultracode/SKILL.md. Actual native model gpt-6-luna, reasoning high, discovered in host controls/cache and used successfully. Six independent packets completed within the requested 72 total ceiling, at most six concurrent helpers. The installed workflow has no named 72 preset; this does not claim 72 completed reviewers. No nontrivial bug candidate was selected, so four-pair adjudication was not triggered.

The foreground call now matches the real seven-argument API. 0, NULL passes validation, adds no stream variables, and retains selected-session credentials, supervision, read-only observation, output, exit and cancellation behavior. The fixture matches that API and checks both empty-environment arguments. Both patches and final changed blobs exactly match the cited master fixes c29254210cf1604d2980dd99556e4fd0baf5bd69 / 83bb316804190cd2919c6477e630bdd2b129bdac; no unrelated files are carried.

I verified round 1 and its completed session record, including six separate helper completions. It is a different session, covered the same refs, and reported no actionable findings. No outstanding change request or review thread exists.

This session independently passed the full adapter script with MSYS2 G++/Node, a positive extracted-route harness, and four generated negative controls: the old five-argument call fails compilation; nonzero count, nonnull array and nonempty environment fail the fixture assertion. SHA-pinned scratch inputs hash to the published blobs. Exact-head PR CI succeeded, including Arch Build PKGBUILD/package contract. Skipped Windows tests and hosted Linux validation are not claimed as executed.

Readiness evidence: open, draft, conflict-free PR; fresh head/base and canary tip unchanged. The allowed branch summary reports protection disabled, required status checks off and empty; the active canary ruleset only prohibits deletion/history rewrites. Other active rules target master/tags. No enforced check requirement is being waived.

Limits: detailed branch-protection GET returned 403 (integration access); stopped without retry. Native Linux broker C/policy/machine-service compilation and installed desktop integration were not repeated on this Windows host; prior cloud results remain implementation evidence. Previously denied local Git inspection was not retried. Conditional draft-to-ready promotion is authorized; the parent retains its own cloud check and merge.

AI-generated COMMENT review under the authorized Bugfix PR Review workflow.