docs(gap): refresh product-technical baseline (74 repos, live census) - #1116
docs(gap): refresh product-technical baseline (74 repos, live census)#1116seonghobae wants to merge 79 commits into
Conversation
…losed review-gate RCA Base revision moved to develop@749511c3. Open-PR count 130 -> 185 (6 days, +55; only #957 landed). Section 5 adds finding (k2): all sampled PRs pass code gates but the three org-owned required reviews (opencode-review, strix, noema-review) fail closed, blocking every PR. Records observed in-flight central repair (noema call_llm timeout branch) as a do-not-duplicate item, and re-scopes P0 #3 to gate remediation as the single top priority. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SoJBAAXwv58S8P4hQBQQAw
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
📝 WalkthroughWalkthroughBandScope의 제품·기술 기준선과 운영 증거 문서를 갱신했습니다. 최신 census, 제품 요구사항, 병합 증거, transport 상태, 영속성 계약, 보안 경계, 품질 및 릴리스 기준을 반영했습니다. ChangesBandScope 기준선 문서
Estimated code review effort: 2 (Simple) | ~15 minutes Merge Risk: 🟡 Moderate · up to The recovery state model currently leaves RecoveryFailed with no documented exit, so a failed restore can dead-end users and encode an incomplete product contract. Merge should wait until the contract defines acknowledgement to NoSource or retry behavior, with regression coverage. 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
… progress Iteration-2 status: gates still fail closed. Central .github landed 8 Noema-reliability fixes (#1477-#1504) plus an active "remove fixed LLM response timeout" branch. Local response: staged merge-ready work behind the closed gates — PR #1116 (this baseline) and PR #1117 (temporal probe promoted from cli hack to api integration, 100% coverage locally). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SoJBAAXwv58S8P4hQBQQAw
|
@opencode-agent Become the sole writer for canonical BandScope branch Validate every finding against fresh live evidence before editing. Current verified contradictions on this exact head include:
Keep the baseline a truth source rather than a blocker narrative: distinguish protected shipped truth, current live evidence, accepted/open work, and planned gaps. If correcting the document surfaces a concrete BandScope executable gap already owned by an issue/PR, link its canonical owner; do not implement that code in this docs lane. Run any repository doc/markdown/contract checks that apply plus |
|
Concurrent recount note for the new exact head |
|
@claude Please refresh the two canonical baseline files on exact current head Also reconcile central-review status from fresh protected-central evidence. Keep the existing Rust core-computation ownership, bounded CPU/accelerator boundary, real-audio acceptance, 100% coverage/docstring/edge-case target, Storybook/Figma/screenshot UX contract, security/operability baseline, and APA 7 traceability intact. Update only truthful volatile evidence and directly contradicted status prose. Run the documentation/baseline contract tests and |
|
Active Player Draft navigation refresh (not shipped baseline truth): protected source remains |
Canonical baseline owner
Single writer for
docs/product-technical-gap-baseline.md. Protected source and exact-head evidence are authoritative; active PR behavior is never promoted into shipped truth here.Current authority
develop@314ddeae7b775a4957594b599358c8255617eb2e.3dbdecf33f4f60d21541af68ec5c7bf22f45b1d3; this metadata update does not claim new source evidence.841e1c9b7329dba6d0ff16daecc009a2c3face0c.8f441881f8d3a5b3fce4ab2bcf934f6fe8ef0397.7c773cdda932855c0e1b9d114c70bb14acc5adea, preserving its 22 unique queue-control files.Current Project Persistence evidence
The intermediate ACL-only gap is now closed further on #970. Renderer-authored
sourceReferenceremains rejected. Resource Admission remains the owner of app-owned source bytes and publication identity; Project Persistence consumes that identity through a typed ACL and native save adapter.Current causal chain after the ordinary #866 adoption:
3a96be63445aba40b52c6a086ff8e4ee0623f466→c2d3541ff3f1a3108adc6224b1dd565fa2a1bd83→b0355a98d03c0e2128f0f1b8b79df9f779514d96: exact path-free identity projection, canonical Resource Admission size policy reuse, exported ACL.6eea76fdb138838d61e8af0d23ea69d99012de21→cb7f4fd956278d1273e6be3f2df367171baadf9e: browser preview fails closed instead of claiming durable Save success without writing bytes.cd3c67de8c9d35355aa30950733a1cf24a5d23fc→ffdac30e63c4faa7264416bbeec8570a0c6543ff: native Save accepts only an explicit optional BandScope project id, performs exact retained-identity lookup, revalidates through the typed ACL and injects v3sourceReferencebefore serialization. Renderer path/digest/artifact/byte evidence and global last-selected-project shortcuts remain forbidden.570894b91f69b0e24c26090309fe5ee5d414f514→979ac4d3a948ab76a01e076fa29bece6161489d6: renderer bridge forwards only{ payload, projectId }.28d94d0e9566030c53370829484f808f6763fbcf→06afcbe030ef1fb8d6b2097be0e40bc4d5c7c03a: mounted local-audio analysis retains its exact project id through Save, scoped to the submitted/result aggregate rather than whatever source is selected later. YouTube replacement and project reopen do not inherit this OS-selected-local publication authority.8f441881f8d3a5b3fce4ab2bcf934f6fe8ef0397: traceability aligned with the current native handoff and restart residual gap.The principal remaining advertised v3 source-persistence gap is now restart re-admission: persisted
sourceReferencemust resolve only the app-owned artifact, re-establish regular/no-link containment, bounded size/SHA-256 and applicable decode/admission, reconstruct fresh bootstrap/native identity, and only then permit #1160 to resolve persistedselectedPlaybackSourceagainst fresh stem availability. Missing preferred stems fail closed to Full mix.Baseline source-currentness
docs/product-technical-gap-baseline.mdat3dbdecf…still names an older #970 semantic snapshot. It does not record the now-completed native retained-identity Save handoff and therefore is not fully code-current for this boundary.Do not promote it as complete. The next product-semantic #1116 source repair must update the Project Persistence/Crash-safe Project rows through #970
8f441881…and move the residual gap to restart re-admission. Because that changes the canonical baseline blob, immediately follow with an ordinary non-force #968 reconciliation preserving all 22 queue-owned files and leavingdocs/product-technical-gap-baseline.mdnon-divergent on #968.Protected gate defect
Fresh branch data still reports protected
develop@314ddeae…with 14 required contexts, including retiredAnalyze (javascript-typescript)andAnalyze (python). #1172 remains the owner of exact context-name migration to the current central CodeQL producer; duplicate local scanning or weaker coverage is not an acceptable workaround.Verification / merge gate
Keep Draft. #1116 source head has not changed in this metadata-only update. #970 exact-head workflows are newly materialized and non-terminal, so no GREEN/Ready/merge claim transfers from predecessors. No self-approval, force-push, destructive rebase, bypass, gate weakening or stale evidence transfer.