Repository navigation
Tracking: Make Aspire CI reliable, actionable, and efficient #18036
Description
[automated]
Status: 🚧 In progress · 👀 Ready for review · ⛔ Blocked · ⏳ Pending · ⏸️ Deferred · ✅ Done
| Last checked | Priorities | Included ownership | Review queue |
|---|---|---|---|
| November 12, 2026 (UTC) | Faster public/internal builds and internal reliability; test-selection maintenance second | Issues/PRs opened by radical; Copilot-opened PRs only when assigned to radical; explicit user-requested exceptions | Global review-ready and active-work counts unverified; review rows below are author-declared complete |
In progress
| Status | Outcome / tracking | Implementation | Next step |
|---|---|---|---|
| ☐ 🚧 Integration and trial decisions | Weekly test-selection drift audit — microsoft/aspire#18036 (inline) | microsoft/aspire#20324 (draft, mergeable) | Resolve credential/reporting decisions; validate the audit before publication. |
| ☐ 🚧 Author conflict/validation work | Platform-specific infrastructure test lanes — microsoft/aspire#21812 | microsoft/aspire#21814 (draft, conflicting) | Resolve conflicts and validate the reconciled head; the author checklist still expects follow-up. |
| ☐ 🚧 CI failure-cause guidance | Improve failure-cause analysis — microsoft/aspire#18036 (inline) | microsoft/aspire#22108 (draft) | User review before any ready-state change. Current scope is guidance/docs, not analyzer code; exact-head workflow validation passed. The description separately reports 47 Windows fake-gh harness failures, not independently reproduced here; no all-tests-pass claim. |
Review requested (author-declared complete)
| Status | Outcome / tracking | Implementation | Next step |
|---|---|---|---|
| ☐ 👀 Reduce full test runs for shared test fixtures | microsoft/aspire#18036 (inline) | microsoft/aspire#21810 | Author checklist declares ready to ship; human review requested. Exact-head CI succeeded after the earlier local feed limitation. No new independent readiness assessment. |
| ☐ 👀 Replace stapler retry JavaScript harness with C# | microsoft/aspire#18036 (inline) | microsoft/aspire#21811 | Author checklist declares ready to ship; human review requested. Exact-head CI succeeded. No new independent readiness assessment. |
Pending work and decisions
| Status | Outcome / tracking | Implementation | Next decision |
|---|---|---|---|
| ☐ 🚧 In progress | Homebrew release PR automation — microsoft/aspire#18036 (inline) | — | Tap draft updated to 89a2b248; native Intel/ARM validation passed, with no run artifacts; CLA passed. Aspire continuation/timeout implementation is committed/pushed at 63fd794f; no Aspire PR yet. Two exact-revision dry-run previews returned expanded YAML. Contained run 3133583 and positive-path contained run 3133587 select respectively source 3115399 (13.6.4) and 3130775 (13.6.5) at that revision. Run 3133583 failed online livecheck before installation; its behavioral/source-artifact jobs passed. Run 3133587 completed/succeeded: PublisherBehavior, SourceArtifact and CanonicalMacOS passed; native install/restore/zap/uninstall passed and nonempty CLI logs were retained. Warning triage verifies 27 scanner alerts (9 high, 11 medium, 7 low), agent-image CVE warnings and SDL no-break baselining; payload attribution and security clearance remain unestablished. Original-release handoff/production continuation and protected live resources remain unvalidated. Hosted handoff validation needs a genuine retained original-release handoff: the contained graph produces none, and the existing producer is in the production original-release graph only. Resolve that prerequisite/approach before continuation validation; no separate test-only producer is approved. Complete review and protected-resource authorization before live enablement. Current-run metadata verifies long retention; two older retained summary artifacts were still downloadable beyond the current 60-day artifact window. Homebrew diagnostic expiry and effective download audience remain unverified. Resolve retention/access and containment checks; no actual release publication, merge or readiness change authorized. |
| ☐ ⏳ Scope decision (investigation complete) | Analyze and auto-rerun validation workflow failures like CI — microsoft/aspire#22119 | — | Decide native analysis scope; review the proposed issue correction. No implementation approved. |
| ☐ ⏳ Prioritized backlog | Sign and publish public CLI acquisition scripts — microsoft/aspire#15933 | — | When prioritized, assess existing signing/publication and whether missing work belongs in the release pipeline; discuss the approach before implementation. |
| ☐ ⏳ Prioritized backlog | Explicit NuGet package signing responsibilities — microsoft/aspire#20584 | — | When prioritized, identify relevant remaining criteria and reconcile existing signing work; discuss the smallest approach before implementation. |
| ☐ ⏳ Approach undecided | Internal native/signing critical path — microsoft/aspire#21305 | radical/aspire#159 (experimental draft) | Six-build experiment inconclusive; decide next approach. |
| ☐ ⛔ Validation and integration | Select artifact producers from consumers — microsoft/aspire#18036 (inline) | microsoft/aspire#20613 (draft) | Establish consumer coverage, integration and practical benefit; fewer runner-minutes do not prove faster feedback. |
| ☐ ⏳ Priority undecided | Consistent daily CLI archive/checksum — microsoft/aspire#20975 | microsoft/aspire#20976 (draft, conflicting) | Decide relevance and install strategy before resolving conflicts and validating platforms. |
| ☐ ⏸️ Low priority | Nuxt validation diagnostics and Node 22 coverage — microsoft/aspire#21146 | — | Choose remaining user-owned work; health-wait recovery is not established by the earlier npm fix. |
Evidence, qualifications, owned backlog and earlier completions
Evidence and limits
| Area | Verified evidence / qualification |
|---|---|
| Ownership | GitHub author and assignee metadata was checked for every linked issue/PR. Each outcome needs its own qualifying tracking issue or implementation PR; the parent link alone does not qualify inline work. Non-owned implementation PRs are excluded even when the tracking issue was opened by radical; user-opened tracking outcomes remain. No implementation is attributed to radical merely because radical opened its issue. The Homebrew release-automation row and microsoft/aspire#20584 are explicit user-requested exceptions; the latter was opened by joperezr and is assigned to radical and joperezr. Other exclusions remain unchanged. |
| Native-hosting analysis approach | Recommendation, not approval: initially analyze failed main pushes and eligible PR runs (including forks and non-main bases), producing only analyzer run summaries/artifacts; no PR, issue or cause-history writes and no auto-reruns. Permission feasibility is established only at compile time in the gh-aw v0.89.21 v6 fixture with explicit job-token pins: no repository-write or actions: write GITHUB_TOKEN permissions; copilot-requests: write remains. V4's optional PAT fallbacks are not constrained by job permissions. Runtime, native end-to-end behavior and v0.89.17 v6 output remain unverified, as do PR identity/freshness and untrusted-evidence handling. Disabled issue reporters leave failures without proactive notifications. This does not complete the requested guarded-rerun outcome. The separate issue correction awaits explicit approval: its cited feed errors recovered before protocol-test failures. |
| Bundle implementation | microsoft/aspire#21304 merged November 7 at 9b8d7c09d72146ead03236e898673c359f6f8afe; microsoft/aspire#21087 is closed. Implementation completion is not measured speedup or pipeline health. |
| Internal signing experiment | Six-build old-head experiment complete and inconclusive: no demonstrated speed, reliability or cost improvement. Representative candidate 3127814 succeeded; baseline 3127889 failed and baseline 3128091 succeeded. Candidate 08f4839 vs old base 97719af, not the final merged-parent revision; preserve the failed run and image/cache caveats. Decide the next approach; no further builds or review readiness implied. The experimental draft is based on the older 97719af revision. |
| Completed bounded main-build watch | Definition 1602 automatic-main build 3129461 and build 3129504 both include the bundle merge and completed literally succeeded: 2/2 scoped qualifying successes, meeting the approved stop threshold. Prepare Installers and Assemble + Publish succeeded. Referenced promotion 3129495 and promotion 3129558 also succeeded. This is not every main build, measured speedup or resolution of the earlier timeout cause. |
| Watch validation limits | PostBuildSign=true skipped in-Assemble NuGet signature verification. No final NuGet cryptographic verification or final-signed npm consumer reruns are claimed. Successful scan tasks do not establish exhaustive coverage. Earlier failures and warnings remain evidence, not retroactively cleared outcomes. |
| Earlier internal main-build observation | Definition 1602 build 3128972 at the merge SHA and build 3129064 at c37dfbed834193aaefb9fd88b5472176dd224c0a both failed: 0/2 inspected qualifying passes, not an estimate of every main build. Seven bundle/CLI producers and fourteen CLI tool/npm verifiers succeeded in each; Prepare Installers and Assemble + Publish skipped. |
| Scan-worker failure signal | Build 3129064 log 606 reports an empty definition version/1970 timestamp; log 617 reports definitions 15,971 minutes old. mdatp definitions update remained pending before worker cancellation after approximately 44–45 minutes. Linux ARM64 MDE warning outcomes are distinct from canceled Linux x64/musl jobs; they do not establish successful scanning. |
| Pre-existing signature | Baseline build 3128882, at source 399e3e6f8984c75dc2778e1329f46958f78f9ad7 two commits before the squash merge, shows the same update wait/worker-timeout pattern. This supports a pre-existing failure signature, not exhaustive exclusion of bundle causality, malware detection or a fully diagnosed root cause. |
| Earlier artifact/runtime coverage | All seven RID archive artifacts were listed in the inspected builds; independent signature verification of downloaded artifacts was not established. Selected native smoke checks passed, others skipped. Successful producer checks do not establish full runtime, antimalware or final-assembly validation. |
| Earlier Homebrew validation | Build 3129480 passed selected-run native artifact download/provenance and publisher behavioral tests. Verified validator command order establishes that its structural gate passed before the SDK prerequisite failed with exit 155. After the working-directory correction, contained rerun 3129492 at a9a4b02c86ec3839f270fa653af3b8482207426c passed the SDK prerequisite and smoke-app creation, then failed at aspire restore exit 6 / child dotnet restore exit 1 (log 111). A verified local isolated-HOME probe with the checksum-matched CLI 13.6.4 archive and matching templates passed CLI restore and a subsequent direct restore on SDK 10.0.400. That was not reproduction of the hosted failure on SDK 10.0.403 or proof of SDK incompatibility. That run did not retain its underlying hosted diagnostic. Both failures and passed paths remain historical evidence; the later contained success does not establish production publication. |
| Earlier Homebrew diagnostic retention | SDK-log retention at 547b48080347b5df45e05fb9474032ca54296739 was exercised by contained run 3129627: notarization passed before restore failed with CLI exit 6 / child SDK exit 1 (log 113). Its homebrew-canonical-macos artifact retained the 979-byte SDK 10.0.403 failure log through cleanup: NuGet.targets(198,5): error MSB4181: The "RestoreTask" task returned false but did not log an error. The subsequent run 3130749 retained NU1105 for all four projects: 'auto' is not a valid version string. This supersedes the earlier missing-diagnostic state. |
| Current Homebrew validation | Primary template comparison verifies the selector rename from VERSION to HOMEBREW_RELEASE_VERSION at 646499c2b0fe3f00cc2c520b32504e96c6e56303. Contained run 3130783 at that revision completed literally succeeded, with HomebrewTestOnly=true, DryRun=true and the same producer 3115399. Its canonical artifact says outcome=validated, CLI 13.6.4, publish=false; log 113 confirms install/version/sidecar, notarization, new/restore, zap preservation and uninstall cleanup passed on SDK 10.0.403. This is contained validation, not full release integration, production publication, review readiness or security clearance. |
| Homebrew authorization hold | After the explicit stop following 3130749, the subsequent commit/push and run 3130783 occurred without specific user authorization, as acknowledged in the corrected execution record. The earlier authorization claim was withdrawn. The user later prospectively resumed bounded work; this did not retroactively authorize that breach. [automated] After the subsequent pause, the user directed continuation of Homebrew work and validation on 2026-11-11, not an explicit 'local only' restriction. At that earlier resumption snapshot, execution was local and no commit, push, hosted run, production action or PR-state change had occurred; this does not describe the later authorized adoption commit/push recorded below. |
| Homebrew publisher adoption | Adoption is committed/pushed to the internal validation ref as e933eb27d006f0c5ce5ff039ebf48380e911e16b, pinned to tap 6221135ffc2bba7f92186154d2ef4cbeaf705ec1; microsoft/homebrew-aspire#36 remains open/draft, not merged or a separate initiative outcome. A nonexecuting preview expanded PublisherBehavior, SourceArtifact and CanonicalMacOS plus inherited SDLSources/AgentlessTag and SDLSources/SDLSources_self with normal System.AccessToken usage; the strict graph gate stopped execution. Neither positive nor controlled-failure validation has executed; the latter awaits a positive pass. Native tap run 42977520789 at the tap pin failed: Intel and ARM passed diagnostic tests/style/checksums but failed livecheck (13.6.4 versus 13.6.5) before install; it is not green. The three ANSI, quoted-header and spaced-value reproductions at old 1aca819 are attributed to the focused handoff; primary threads were read directly, exact probes not independently rerun. The pushed fix passed 9 runs/57 assertions on Linux per its handoff. Existing artifacts 13827651689/13827094122 at the old pin and 13836047667/13836327608 at the new pin each contained only one [] file; no credential found in those four, no broader historical disclosure conclusion. Uploads using unsafe tap 1aca819 remain prohibited; corrected 6221135 contained validation remains subject to the preview graph gate, with a passed, containment-verified positive run required before controlled-failure execution. Producer 3130775 at b33d222 matches v13.6.5 and its artifact passed local artifact-only/consistency checks, not a full install, hosted publisher upload or unified-producer provenance proof. Local real-starter baseline restore passed and forced VERSION=auto failure retained NU1105/MSB4181; full adoption install/upload and production lifecycle remain unverified; actual release publication is not authorized. Earlier execution and stop-boundary history remain recorded separately and in the private note. The raw-diagnostics tap revision is now committed/pushed as 89a2b248e4c7ab883d072ecb2d10874e04c2d6b1, integrating current base fca1b2e641537711407b019299f1d47fd87180f5 through merge 95ed04999bf4a2a397ae9061a6ac7ceaac25f6bc; it removes the redaction collector/public uploader and retains raw traces plus failure assets in the validator diagnostics directory. Complete patches/commit messages were shown; the earlier approval wait was superseded by the user's November 11 evening authorization recorded below. Published tap draft is now 89a2b248; Aspire fork and internal branch now point to 63fd794fdb9fb8f00dc1ab6159a21b8a22f752db, and no Aspire PR exists. Native tap run 43122495774 completed successfully at that exact head: both macOS 15 Intel and ARM jobs (Homebrew 7.0.9) passed. CLA passed. The completed-run artifact inventory is empty (0 artifacts); this is not raw-upload or confidentiality evidence. Separate Aspire integration and production lifecycle validation remain unverified. Published Aspire pins reference 89a2b248; the 15-path commit includes exact-source handoff, separate post-publication continuation, dry-run/protected live paths and process-group timeout cleanup. The earlier local 15-path snapshot matched its saved tested preview; that observation does not validate every subsequent final-commit change. Two exact-revision, explicit dry-run preview requests (contained and continuation) returned expanded YAML with response ID -1 and no execution result; these are nonexecuting compilation receipts, not hosted runs. The earlier previews created no run; subsequently contained run 3133583 started at exact 63fd794f, with DryRun=true/HomebrewTestOnly=true/HomebrewPublishOnly=false. Live resource metadata resolves source 3115399 (definition 1602, build 20261023.11, release 13.6.4), not latest. At inspection PublisherBehavior succeeded, with behavioral log 29 reporting 44 tests/OK; inherited SDL and SourceArtifact jobs succeeded, including downloaded-cask identity/selected-run provenance and transfer. The overall run subsequently completed/failed: validation log 111 reports Version '13.6.4' differs from '13.6.5' retrieved by livecheck, then cask checks failed; installation was not attempted (Bash exit 1). The failure-path publish task succeeded and artifact inventory includes homebrew-canonical-macos; its contents were not inspected here. Passed source-artifact contracts do not establish native installation or overall success. Separately, positive-path contained run 3133587 completed/succeeded at 00:50:33 UTC at the same exact head and dry-run/test-only parameters; its PublisherBehavior, SourceArtifact, CanonicalMacOS and inherited SDL jobs all succeeded. Downloaded-cask identity/provenance, exact-source retention/transfer and the full source-cask validation task succeeded. Direct native task evidence verifies install, project creation/restore, zap preservation and uninstall cleanup. Retained internal artifact metadata reports validated/version 13.6.5/publish=false with the same selected-source provenance; its CLI-new/CLI-restore logs are nonempty. Raw diagnostic contents were not inspected or published by this reconciliation. This is successful hosted contained execution, not original-release handoff/continuation, protected-resource behavior or production publication. Primary component scan log 116 reports 27 alerts (9 high, 11 medium, 7 low); source-agent pre-job log 70 warns about an installed-image CVE, and SDL log 83 reports missing baselining/no-break execution. Earlier release run 3132189 also reports that image CVE and its component scan shares 15 exact CVE/component-version/severity tuples. Checkout and scan-job contexts differ, so neither the other alerts nor Homebrew payload impact can be attributed. This completes bounded warning-source/comparison triage, not remediation, demonstrated product exploitability or security clearance. Its API-resolved source is 3130775 / 20261109.7, independently verified as definition 1602, succeeded, tagged 13.6.5. The successful contained run validates this selected source; it still does not prove the original production release's source, and earlier local artifact checks remain local. The historical 13.6.4 online-livecheck failure is now observed in run 3133583; run 3133587 now has a successful contained validation result, without establishing production readiness. Independent review and hosted validation remain unfinished. A saved command-stderr receipt records a separate preview-only live continuation failing because the protected publisher environment could not be found or authorized (absence versus authorization is unresolved); no live continuation run or resource provisioning is established. Earlier saved local results: 43 Python behavioral cases and 388 focused infrastructure cases passed (the latter includes the Python wrapper, not independent additional coverage). These do not establish native failure cleanup, hosted continuation or production readiness. Saved read-only inventory reports the protected publisher environment and credential group absent; provisioning/authorization and hosted graph/handoff validation remain outstanding. Earlier published pins and failed runs above remain historical evidence. Earlier saved fixture results: tap 25 runs/101 assertions/0 failures; publisher then 30 tests/OK, including draft/prerelease/mismatched-tag release guards. Each guard's disposable mutation failed its corresponding subtest; these are local fixtures, not native Homebrew or hosted integration. Containment checks and production-publication boundaries remain. A local actual-validator probe with real CLI 13.6.4/SDK 10.0.400 retained both success logs and, on forced restore exit 6, four byte-identical project-relative assets. Homebrew/bootstrap/notarization were substituted. Durable validator cleanup is unverified: configuration reappeared after initial cleanup, cause unknown; the overall probe did not pass. This does not establish full install or hosted integration; access/retention and containment checks remain. A shortened-timeout probe of the actual publisher/modified validator with fake tools reproduced SIGKILL bypassing EXIT-based asset retention/uninstall/workspace cleanup and leaving a descendant; direct logs survived. Exact owned processes/workspace were subsequently cleaned. Separate real-CLI evidence found surviving MSBuild node-reuse workers; the configuration writer remains unknown. These lifecycle limits predate the pending raw revision and are not hosted/full-install evidence; access/retention and containment constraints remain; the earlier no-queue hold was superseded only for authorized targeted validation. At the earlier inspected proposal, production wiring was absent: it selected only HomebrewTestOnly contained validation; completion of new wiring is not established here. The release process requires release-manager publication after automation/verification; the existing pipeline dispatches draft-release tasks and uploads assets before that step. The publisher requires an already-published stable release. On November 11 evening, the user approved the presented tap integration/commit/push, Aspire repin, necessary release wiring/timeout cleanup and targeted hosted validation. The separate publication/ESRP remark was retracted; actual release publication, merge and PR readiness remain unauthorized. This approval is not evidence that changes or hosted runs have completed; raw-artifact access/retention and containment checks remain prerequisites. Handoff validation prerequisite: current source and expanded previews place the only actual handoff producer in the stable original-release graph with DryRun=false; contained test-only validation produces no handoff. Even a non-publishing continuation must download and resolve a genuine retained original-release handoff. Preview expansion is not transfer evidence; a usable handoff remains unestablished. Resolve that prerequisite or approve a distinct safe test-producer approach before hosted continuation validation; no separate test-only producer or production release is authorized by this reconciliation. Current diagnostic-retention risk: successful run 3133587 is marked keep-forever and retained-by-release with long-lived leases; failed run 3133583 is also retained-by-release with a long-lived lease. Project artifact/run defaults are 60 days. Independently reproduced downloads of NpmValidationSummary from retained runs 3063758 (finished September 2) and 3062416 (September 1) show other pipeline artifacts remain downloadable beyond that current window; both runs are retained-by-release, not keep-forever. This proves access only for the current authenticated identity and those summary artifacts, not historical policy, Homebrew logs, other users' access or guaranteed expiry. Contents were not read or exposed; temporary downloads were removed. Actual Homebrew diagnostic-artifact expiry and effective audience remain unverified; no disclosure is claimed. |
| Local exact-SDK comparison | An existing, verified SDK 10.0.403 probe with isolated HOME/package cache and skip-restore scaffolding passed CLI restore; the subsequent direct restore also passed but shares that cache. This is not a hosted reproduction, full Homebrew validation or evidence for an SDK/cache fix. No new probe was performed for this reconciliation. |
| Local release-selector reproduction | Existing macOS SDK 10.0.400 script/log/assets receipts show minimal-project restore fails with VERSION=auto (exit 1, NU1105), while HOMEBREW_RELEASE_VERSION=auto succeeds (exit 0, no assets errors). The probe uses separate temporary projects/caches, not a byte-identical hosted replay. Together with the primary template and hosted error/success receipts, this supports a selector/project-version collision, not an SDK/feed/cache fix. No reproduction was executed for this reconciliation. |
| Review readiness | #21810 and #21811 explicitly check author-complete/ready-to-ship in their current descriptions and request human review; that is the basis for the author-declared review rows, not non-draft status or green checks alone. Global review-ready and actively worked PR counts remain unverified; do not infer capacity from this board or inactive drafts. Across projects/repositories, at most five user PRs may have author work complete with only review remaining; ask the user to prioritize before exceeding five. Separately, target three actively worked user PRs, allowing up to five only when the review-ready backlog is light; ask if extra active work does not clearly fit. These are separate limits, not a combined cap or a numerical definition of light backlog. Investigations without a PR can continue. No readiness change, new implementation or hosted run is authorized by this report. |
| Signing follow-ups | Both issues remain open and are retained as prioritized backlog, not an immediate implementation commitment. Acquisition-script assessment covers public scripts, not PR dogfood scripts; a separate pipeline is not a settled requirement. NuGet assessment must distinguish current coverage from relevant remaining work and reconcile the existing follow-up before duplication. Discuss both approaches before implementation. Neither outcome is claimed complete; official integration remains unestablished. |
| Measurement | Selective execution is deployed, but program-wide PR wall-time improvement is not established. Comparable baseline/changed elapsed time, runner cost, variance, regressions and preserved consumers are required before claiming improvement. Preserve initial failures when reporting recovery. |
| Reporting | Rolling CI Health — microsoft/aspire#18231 remains open. Weekly progress reports are on request and distinguish shipped impact from pending work. Daily Pulse — microsoft/aspire#18232 is closed; microsoft/aspire#20338 is a historical audited baseline, not a current inventory. |
| Scope | Infrastructure, measurement, prevention, CI tooling and installer integrity remain here. Quarantine is excluded. Ownership inclusion does not change separately deferred project scope or authorize implementation. |
Retained / deferred owned outcomes
| Status | Outcome / tracking | Implementation | Next decision |
|---|---|---|---|
| ☐ ⏸️ Deferred | Remove dependency waits — microsoft/aspire#21015 | — | Decide whether/when to resume; the closed prior implementation is not active. |
| ☐ ⏳ Priority undecided | Public-channel CLI validation — microsoft/aspire#21271 | — | Establish publish-lane coverage before selecting an uncovered installation/version check. |
| ☐ ⏳ Priority undecided | Generated-workflow compiler/lint blocker — microsoft/aspire#21031 | — | Recheck reproducible compiler/output lint behavior before resuming. |
| ☐ ⏳ Priority undecided | Artifact invariants — microsoft/aspire#21164 | — | Inventory package-content and post-signing protection; candidate gaps are not confirmed omissions. |
| ☐ ⏳ Priority undecided | Invalid selector trigger maps — microsoft/aspire#21165 | — | Establish malformed-rule/reference behavior and production-path negative coverage. |
| ☐ ⏳ Priority undecided | Deployment workflow PR validation — microsoft/aspire#20491 | — | Define useful dry runs without credential exposure or mutation. |
| ☐ ⏳ Priority undecided | Embedded-script behavioral tests — microsoft/aspire#20657 | — | Extract only what is needed for focused executable coverage. |
| ☐ ⏸️ Retained | Duplicate failure reconciliation — microsoft/aspire#19578 | — | Reassess semantic reuse and race/replay gaps against the merged typed analyzer. |
| ☐ ⏸️ Retained | Unusual-path selector diagnostics — microsoft/aspire#19786 | — | Establish current defects while preserving conservative run-all fallback. |
| ☐ ⏸️ Draft retained | API/ATS generation PR validation — microsoft/aspire#18036 (inline) | microsoft/aspire#20498 (draft, conflicting) | Decide whether to resume; then resolve conflicts, establish release-branch routing and validate the final head. |
| ☐ ⏸️ Retained | Benchmark project routing — microsoft/aspire#19986 | — | Confirm routing outside the solution graph without weakening fallback. |
| ☐ ⏸️ Draft retained | Structured results despite oversized diagnostics — microsoft/aspire#21469 | microsoft/aspire#21732 (draft) | Confirm remaining author work, fallback integration and final validation with the existing owner; draft/checklist status does not establish active work or readiness. |
| ☐ ⏸️ Draft retained | Analyze setup/build failures without test reports — microsoft/aspire#21779 | microsoft/aspire#21782 (draft) | Confirm the metadata-dependent missing-report contract and remaining integration/readiness work; do not launch a duplicate repair. |
| ☐ ⏸️ Draft retained | Preserve unresolved diagnoses — microsoft/aspire#21778 | microsoft/aspire#21785 (draft) | Reconcile integration with the cause-guidance draft before new work; retain this separate outcome without claiming current author activity. |
Earlier completed owned implementations
| Outcome / tracking | Implementation | Completed | Qualification |
|---|---|---|---|
| Typed failure analyzer — microsoft/aspire#20332 | microsoft/aspire#21347 | 2026-11-03 | Issue closed; semantic duplicate reuse remains separate. |
| Node 22 Nuxt npm failure — microsoft/aspire#21146 | microsoft/aspire#21147 | 2026-10-20 | Remaining diagnostics/recovery tracked above; issue stays open. |
Recently completed
| Status | Outcome / tracking | Implementation | Completed |
|---|---|---|---|
| ☑ ✅ Done | Shared bundle build and component verification — microsoft/aspire#21087 | microsoft/aspire#21304 | 2026-11-07; implementation merged, pipeline health assessed separately |
| ☑ ✅ Done | Bounded post-merge native-bundle watch — microsoft/aspire#18036 (inline) | microsoft/aspire#21304 (merged) | 2026-11-07 (local); approved 2/2 qualifying automatic-main success threshold reached |
| ☑ ✅ Done | Current-main infrastructure retry policy — microsoft/aspire#20340 | microsoft/aspire#20345 | 2026-11-07; intentional one-retry infrastructure scope accepted, issue closed (PR merged November 5) |
| ☑ ✅ Tracking closed | Windows confirmation retry timing — microsoft/aspire#20968 | — | 2026-11-06; user-opened issue closed, not a user-authored implementation claim |
Activity
radical commentedon Aug 29, 2026
cc Jose Perez Rodriguez (@joperezr)
18 remaining items
tests: Add guidance around creating Infrastructure tests (#20612)
Avoid Windows batch reparsing of parsed workflow arguments
fix(ci): Preserve native Windows build logs
refactor(ci): Keep single-consumer workflow logic inline