RN 0.82.1 certification requires a fmt 12.1.0 backport

Finding

Severity: Medium. Unmodified RN 0.82 fails fmt compilation on the tested Apple toolchain.

The subsequent application evaluation passed with an explicit fmt 12.1.0 patch.
See the evaluation comment for exact commits and commands.

The operator approved RN 0.82.1 minimum adoption for 0.4.0-rc.2.
Implementation and release certification remain incomplete.

Initial source and evaluation scope

Synchro source: 7bc2f7a34b619099edd6cb0645e44de19fca1e7a.
The repository worktree was clean before and after evaluation.
An isolated directory contained source archived from that commit.
Evaluation dependencies were React Native 0.82.1, React 19.1.1, Codegen 0.82.1, and TypeScript 5.9.2.
The npm registry lists only stable RN 0.82.0 and 0.82.1.
The available compiler was Xcode 27.0, build 27A266a.
No support declaration or production dependency changed.

Initial executable evidence

make lint-rn082 passed for the production TypeScript source.
make codegen-rn082 passed and generated Android and iOS native declarations.
make compile-fmt11 failed with five consteval errors.
make compile-fmt12 passed with the same compiler and flags.

The compilation command was:

xcrun --sdk iphonesimulator clang++ -std=c++20 -arch arm64 -mios-simulator-version-min=16.0 -I fmt-11/include -c fmt-11/src/format.cc -o fmt11.o

The first failure was:

fmt-11/include/fmt/format-inl.h:59:24: error: call to consteval function ... is not a constant expression

RN 0.82.1 publishes fmt 11.0.2 and requires C++20.
RN 0.83.5 publishes fmt 12.1.0.
The fmt 12.1.0 check isolates the upstream dependency failure.
It does not establish a working RN 0.82 application.

Full native compilation, installed application execution, and packaged lifecycle checks did not run.
These probes are evaluation evidence, not release certification.

Upstream evidence

RN 0.82 Xcode 26.4 fix request.
Upstream closed that request because RN 0.82 reached end of support.
Merged fmt 12.1.0 correction.

The local probe reproduced the reported failure under Xcode 27.0.
It did not test Xcode 26.6 directly.

Required adoption changes

Review RN and React peer ranges in clients/react-native/package.json:88.
RN 0.82.1 bundles React 19.1.1, which the current React peer range excludes.
Add explicit RN minimum cells alongside current cells in conformance/support-matrix.json:133.
Update the fixed support tuples in conformance/internal/contract/policy.go:43.
Parameterize the consumer version pinned in verification/consumers/react-native/test-consumer.sh:83.
Parameterize the upgrade consumer version in verification/consumers/upgrade/react-native/build-app.sh:30.
Update CI, release environment records, and support documentation.
Bind the minimum cells to the existing authoritative semantic runners.

Applicable existing cells are SUP-RN-IOS-CURRENT-001 and SUP-RN-ANDROID-CURRENT-001.
Applicable release gates are Candidate and Package in RELEASE.md.

Updated recommendation and remaining work

Use RN 0.82.1 as the approved minimum for certification.
Require the reproducible fmt 12.1.0 patch for RN 0.82 iOS consumers on current Xcode.
The consumer application must apply this patch to its RN dependency.
The Synchro package cannot enforce a consumer application's package-manager patch.
Do not describe patched RN as compatibility with the unmodified published RN package.

The initial recommendation required an upstream release or an older toolchain.
The later complete application builds support a third path: an explicit, tested dependency backport.
The support declaration and full release certification remain incomplete.

Acceptance criteria

Record the approved RN version and Apple-toolchain scope.
Demonstrate clean native builds from the exact RN version and declared fmt patch.
Verify the packaged bridge on both platforms with real SQLite and the extension-backed adapter.
Run the existing authored semantic corpus with structured results and zero skips.
Verify packaged install, connect, push, pull, process termination, and resume.
Verify each declared minimum OS track and current track.
Keep package peer ranges, support cells, version records, and documentation consistent.
Preserve this failed compiler result with final evidence.

RN 0.82.1 application evaluation

The user authorized validation of fmt 12.1.0 before a support-policy decision.
The evaluation used a separate Git worktree.
The main worktree remains clean at 7bc2f7a34b619099edd6cb0645e44de19fca1e7a.
No declared support cell changed.

Complete application builds

Clean evaluation commit: deee1414646c6e8c745e8a5d23d28c7b18cab1a2.

CocoaPods resolved React Native 0.82.1 and fmt 12.1.0.
make test-rn-android-parity passed.
make -f evaluation.mk eval-rn082-android-build passed.
The Android Release APK includes the native bridge, generated code, and Hermes bundle for arm64-v8a.
make -f evaluation.mk eval-rn082-ios-build passed on Xcode 27.0, build 27A266a.
The iOS Release simulator application contains arm64 and x86_64 slices and a Hermes bundle.
Both iOS slices compiled fmt, Folly, RN, Synchro, and GRDB.

The source change comprises three version entries across RN's two iOS podspecs.
It changes the fmt version and Git tag to 12.1.0.
It changes the RCT-Folly fmt constraint to 12.1.0.
No Synchro production source changed.
Android used its published RN artifacts and needed no CocoaPods patch.

Reproducible patch and unit checks

Clean initial unit-check commit: 1f7df534eac695b4a95f1b312af5f5e96e054b39.

The existing Yarn 4 patch mechanism now applies the dependency correction.
Both workspace manifests reference the checked-in patch.
Both installed RN copies contain the fmt 12.1.0 correction.

make lint-rn test-rn-unit passed.
The structured Jest parser recorded 239 tests in 10 suites, with zero failures and zero skips.
Lint recorded zero errors and nine warnings in unchanged source files.

The initial unit attempt failed before test execution because RN 0.82's Jest setup requires Prettier.
The RN 0.83 setup guards that optional dependency.
The evaluation adds Prettier 3.6.2 as a test dependency.
No assertion or test selection changed.

Native test harness failure

The initial extra iOS transaction attempt used Release mode without Swift testability.
It failed at @testable import SynchroReactNative and executed zero tests.
The structured parser rejected the zero-test result.
This evaluation harness differed from the supported Debug test target.
The correction enables ENABLE_TESTABILITY=YES for this test build only.
It also selects the active simulator architecture for test execution.
No production source or test assertion changed.
The failed log and xcresult remain available.

Native execution results

Clean iOS 27 attempt: 44755f0c99e2f8e1c944c00079907d5568f83821.
The corrected test target compiled successfully.
The generated CocoaPods test host crashed during UIKit's scene lifecycle check before test execution.
The structured result parser rejected the failed run.
This separate fixture defect is recorded in #304.

Clean iOS 18.1 evaluation: 8965675747961ec4defde1f4a2a3c6a304b2b27d.
The following command passed:

make -f evaluation.mk eval-rn082-ios-transactions \
  RN_IOS_TEST_DESTINATION='platform=iOS Simulator,id=BD7DCDC3-2E39-4C1B-A00A-487978B1965B' \
  RN_EVAL_IOS_RESULT=ios/build/eval-rn082-transactions-ios18.xcresult

The structured xcresult parser recorded 11 passed tests with zero skips.
These tests exercise the native bridge transaction target with real SQLite.
This diagnostic older-runtime result does not satisfy the current iOS support cell.
It does not prove the iOS 16 minimum cell.

The full evaluation remains on local branch evaluation/rn-082-fmt-121.
The main branch and declared support policy remain unchanged.

The final clean commit is 8965675747961ec4defde1f4a2a3c6a304b2b27d.
make -f evaluation.mk eval-rn082-install lint-rn test-rn-unit passed on that commit.
This final run confirms the immutable patch install, lint, and all 239 unit tests.

Remaining certification scope

These checks establish a working dependency backport for the tested application.
They do not certify packaged consumers or full synchronization behavior.
The extension-backed semantic corpus and packaged lifecycle checks remain required.
Minimum operating-system execution also remains required.
RN 0.82.0 was not tested.

Proposed minimum: RN 0.82.1 with the declared iOS fmt patch.
Use exact version records and an upper bound at the highest certified RN minor.
Run full semantic proof on minimum and current RN versions for both platforms.
Run packaged install, connect, push, pull, process termination, and resume for every supported minor.
Add minimum operating-system bridge cells rather than relying only on native SDK coverage.

The application owns its RN patch.
Publishing the Synchro library does not apply this patch to downstream applications.
The support policy must state this consumer prerequisite.

Local logs are retained in /tmp/synchro-rn082-validation.Idnu2R-logs.
The original fmt 11.0.2 failure remains in the issue history.
The first iOS build emitted four Swift diagnostics stating exit code zero with no further output.
The build returned success and produced both native slices.
Those diagnostics remain in the retained build log.

Decision

The operator approved RN 0.82.1 minimum adoption and another release candidate.
The candidate version is 0.4.0-rc.2.
RN 0.82.1 uses the explicit upstream fmt 12.1.0 iOS backport.
Existing RN 0.83 support remains required.
Consumer documentation must explain the application-owned RN patch.

Required validation

Complete local Mac and Linux desktop validation before CI starts.
Use GPT-6.1-Sol workers and independent Astra release review.
Keep the existing semantic corpus, assertions, and structured zero-skip checks.
Bind support evidence to the exact clean source commit and executed environment.
Resolve #304 before current iOS certification.

The isolated implementation branch starts at 7bc2f7a34b619099edd6cb0645e44de19fca1e7a.
No implementation or release gate is complete.
The existing assignment candidate in #300 requires a separate integration scope decision.

The next candidate remains 0.4.0-rc.2.

Support decision

Add required cells SUP-RN-IOS-MIN-001 and SUP-RN-ANDROID-MIN-001.
Both cells select RN 0.82.1 on the current stable platform.
Retain the current RN cells and select RN 0.83.10 for their validation.
Both runtimes execute the same authored scenario corpus.
Keep all existing scenario assertions, fault plans, and current performance obligations.
Set peer ranges to RN >=0.82.1 <0.84.0 and React >=19.1.1 <20.0.0.
Use React 19.1.1 with the minimum RN fixture.

Dependency input

The RN 0.82.1 fixture uses the exact fmt 12.1.0 backport.
The patch changes the fmt pod version, source tag, and RCT-Folly dependency.
Its upstream commit is faeef2b90a56633ad44289b994d31e7ce590b145.
Its SHA-256 is da0bcf392ec9b086fab5f540a1ead36684ad7675db2e453d7a83e3ffd1524011.

The consumer application must retain this patch before CocoaPods installation.
Installing the Synchro package does not patch a consumer application automatically.
The documentation states this requirement.

Open validation work

Astra found an incomplete positive validator fixture and an ineffective fallback control.
Sol corrected both source gaps without reducing the support requirement.
Independent source review and executable acceptance remain separate.

Published Synchro 0.3.1 requires RN >=0.83.0 <0.84.0.
Its upgrade proof cannot install that predecessor on RN 0.82.1 with valid dependencies.
The correct upgrade path remains under review.
No dependency checks will be bypassed.

The desktop has the required Android images and usable KVM acceleration.
Application execution and the complete local test gates remain pending.
The iOS 16 minimum gate still needs an accessible compatible Mac.
The release integration scope also remains pending.

No CI, push, pull request, or release publication has started.
No candidate certification or issue closure is claimed.

RC.2 preparation now uses 0.4.0-rc.2.
The tag will be v0.4.0-rc.2.
No CI, push, pull request, tag, or publication has started.

Source preparation

Both runtime locks are committed in b855f6dcc49d529d71c77b921b48930743214398.
Minimum installation measures RN 0.82.1 and React 19.1.1 from both workspace and application resolution.
The installed minimum pod graph contains fmt 12.1.0 and Folly's matching fmt dependency.
Normal installation keeps peer checks enabled.
The three peer metadata warnings also occur in the original RN 0.83.0 checkout at 7bc2f7a34b619099edd6cb0645e44de19fca1e7a.
The minimum upgrade path uses the supported RN 0.83.10 predecessor, then the RN 0.82.1 candidate.
The published predecessor requires RN 0.83.x. No peer bypass permits a minimum predecessor installation.
Native upgrade execution and retained-state evidence remain pending.

Executed focused validation

These commands passed on clean commit b4f309e9c710ece462a236959a821e7ee7b8691f with Node 22.22.0:

make test-rn-runtime-helpers
make test-conformance-contract
make test-conformance-drivers
make test-conformance-scenarios
make test-packaged-smoke-structure
make test-release-artifacts
make test-version-contract
make version-check
make check-released-update-scripts

Structured results report zero failures and zero skips.
The runtime helper runs nine tests.
Contract, driver, and scenario results report 162, 786, and 357 tests respectively.
Packaged smoke structure and release artifact controls report 18 and 32 tests respectively.
The version contract runs one test.
The released update check confirms two immutable update scripts.

Minimum source preparation, lint, and unit execution also passed on that clean commit:

make rn-source-fixture lint-rn test-rn-unit rn-ios-pods RN_RUNTIME_VERSION=0.82.1 RN_WORKSPACE_DIR=<owned-minimum-workspace>

The Jest result contains 239 passed tests, zero failures, zero pending tests, and zero todo tests.
CocoaPods installation completed.
These results do not establish native certification or final release signoff.

Demonstrated negative control

The full make test-conformance-drivers gate passed on clean baseline d27f7d5135937a415082842a298565769f8e2afc.
The isolated clean mutant 173b28f6cd882ba803dc6fb249a372fff31165ea permits invalid explicit application paths to select the valid default.
The same full gate fails the empty, relative, and missing path controls.
No production or test assertion was weakened.

Retained failures and corrections

Candidate Make used unsupported combined override export syntax. The source now uses supported assignments and explicit recursive workspace propagation.
A Candidate dry run executed an existing recursive shell recipe. Immutable installation failed before lock preparation.
Lock preparation first used an invalid Yarn mode. The mode is corrected to skip-build for non-gate preparation only.
The example initially resolved the patch beneath its own workspace. Both manifests now use Yarn's project-root patch selector.
The scenario repository fixture omitted the two minimum targets. The fixture now includes them and keeps its negative control.
The packaged structure fixture expected seven cells. It now requires the approved nine cells.
The first version contract run lacked documentation dependencies. make verify-contract installed and verified them.
Git archive converted the Windows Gradle script to CRLF. Fixture copying now preserves the committed blob bytes.

All failed logs remain in the owned local release evidence directory.

Remaining source review and release blockers

Astra reported two additional source binding defects:

RNR-02: independently overridden derived Make paths can separate verification from execution.
RNR-03: fixture verification does not reject added platform-specific application source.

The implementation worker is preparing bounded corrections in the existing proof home.
Astra must review the corrected source before complete platform validation and release acceptance.
The release scope decision for issue #300 remains pending.
An accessible iOS 16.4 and Xcode 16.4 host remains required for minimum native certification.
The local host cannot provide that cell.

RC.2 local validation update

The committed version is 0.4.0-rc.2.
The planned tag is v0.4.0-rc.2.
No tag, CI dispatch, push, or publication occurred from this task.

Clean integrated source: d205f479e1afeaa3adc759b0f830fcd98c8ccd9c.

Completed evidence

make ci-source-quality passed on the Mac and desktop 10.0.0.86 on that exact clean source.
Both complete structured runs had zero failed and skipped tests.
Their zero-match negative control rejected an empty selection as required by Makefile:562-567.
Current RN unit validation passed 239 tests in 10 suites on both hosts.
Kotlin unit validation passed 718 tests in 46 suites on both hosts.
make lint-rn test-rn-unit passed for the exact minimum fixture on the same clean source, with 239 tests in 10 suites.
make test-rn-bridge-transactions-ios passed all 11 tests for RN 0.82.1 on iOS 27 and iOS 18.1.
That minimum native build used fmt 12.1.0 with Xcode 27.
The same complete iOS 27 gate passed all 11 tests for RN 0.83.10.
Separate make rn-runtime-check calls passed after the minimum and current native builds.
Independent review accepted all three RN source corrections and the descriptor cleanup correction.
The conflicting-path mutant and added-source mutant failed their unchanged helper assertions.
The final journal-cleanup mutant failed the unchanged descriptor comparison, with 14 descriptors observed and 7 expected.
Issue Soak journal leak gate fails while listing macOS descriptors #307 now has complete acceptance evidence and is closed.

The supported iOS gates used owned simulator UDIDs.
Their complete result bundles remain in .ignore/release-rc2/evidence/.
The exact source, lock, installed RN, installed React, and fmt patch identities remain in the fixture runtime records.

Retained failures and active work

The first desktop source-quality attempt failed because the copied checkout lacked published release tags.
Fetching the published tags corrected that prerequisite.
The complete desktop gate then passed on the integrated source above.

The first RN 0.83.10 iOS build failed because the Mac disk became full.
Its structured result contained zero tests and was rejected.
The primary retained that result and removed obsolete task outputs.
The complete gate then passed after disk cleanup.

The first Android bridge attempt built its APK but could not install it into the full 800M emulator data partition.
The corrected owned emulator has 2 GiB RAM and an 8 GiB data partition.
Its measured guest is Android 17, API 37, REL, x86_64.
The complete retry ran 16 tests, with 7 failures and zero skips.
Those failures report a missing app database directory during initialization.
The root cause remains under investigation.
No test assertion or selection changed.

Complete Android failure logs and XML remain retained.
The desktop broad integration mutation gate is running from the exact source above for #305.
The required iOS 16.4 native gate still needs a compatible host.
Full scenario, upgrade, package, security, and final release signoffs remain pending.
These completed results do not establish complete RN certification or release acceptance.

Local preparation uses version 0.4.0-rc.2 and intended tag v0.4.0-rc.2.
No CI, tag, or publication started for this change.

Additional executable evidence ran on clean commit d205f479e1afeaa3adc759b0f830fcd98c8ccd9c:

RN 0.82.1 native iOS bridge: all 11 tests passed on iOS 27.0 and iOS 18.1.
RN 0.83.10 native iOS bridge: all 11 tests passed on both runtimes.
All four runs used Xcode 27.0 and structured xcresult parsing with zero failures and skips.
Separate post-run runtime checks passed for both RN profiles.
make test-swift-unit passed all 351 tests through structured xcresult parsing.
Both complete Mac and desktop source-quality gates passed, as recorded in the previous comment.

The first current-RN iOS run failed because the Mac disk became full.
The structured parser rejected its zero-test result.
The failed bundle and log remain with the later passing results.
Only task-owned compiler outputs were removed before the corrected run.

Android diagnosis found that ContextImpl attempted database directory creation and received mkdir failed: ENOENT.
The saved retry also records disconnected external FUSE storage before instrumentation.
The evidence does not identify the unavailable parent component or prove a Synchro initialization defect.
The complete retry remains failed with 7 failures among 16 tests.
No SDK directory workaround was added.
The failed installation, retry XML, and per-test logcats remain retained.

The desktop also reproduced #305 after 21 completed mutation categories.
Its selected seed-portability assertion passed, but parent setup expired clients that unselected callbacks did not create.
Commit 3d95bc20f37f18af00e665ae0952b1439aa503c4 moves those existing setup calls before their unchanged assertions.
It also retains scheduled mutation failure diagnostics.
Independent source review and complete executable validation are in progress.

RN minimum certification remains incomplete.
The required iOS 16.4 host, Android storage cause, complete semantic corpus, packaged lifecycle gates, and final signoffs remain pending.

Complete Mac Swift Candidate validation passed on clean 0110c472cb75cd05f7b618a9939db80107eb7916.
The source SHA remained unchanged and the worktree remained clean.

Command: make ci-candidate-swift with the owned PostgreSQL fixture and supported artifact paths.
Actual host: macOS 27.0.1, Xcode 27.0, build 27A266a, PostgreSQL 18.3.

Warm-connect passed its exact structured target.
The complete authored Swift corpus passed 14 scenario cases and its parent, with 15 structured tests.
Swift integration passed 381 XCTest cases.
Canonical seed verification passed its exact structured target.
Swift schema identity passed one XCTest case.
Kotlin schema identity passed one JUnit case.
The packaged Swift consumer resolved the selected local artifact and executed successfully.
The retained-intent predecessor upgrade passed its exact structured target.
Go evidence contained 21 passing test events and zero failures or skips.
The complete Make target exited zero.
Both fixture cleanup targets exited zero.

The earlier missing-PG-binary environment failure remains retained.
The correction supplies SYNCHRO_CONFORMANCE_PG18_BINDIR from the actual PostgreSQL18 binary directory.

Separately, the corrected short-path Mac PostgreSQL gate passed on clean 3d95bc20f37f18af00e665ae0952b1439aa503c4.
Command: make test-rust-pg PGRX_TARGET_DIR=/private/tmp/synchro-rc2-pgrx-20261004.
It executed 430 tests with zero failures, ignored tests, or filtered tests.
The failed long-path attempt remains retained with its measured socket-length cause.

Retained evidence includes ci-candidate-swift-0110c472.log, its source/status/exit records, and copied integration and schema xcresult bundles.
PostgreSQL evidence includes rust-pg-short-path-3d95bc20.log and its source/status/exit records.

The required iOS16.4/Xcode16.4 package cell remains unavailable locally.
Current-platform results do not replace that minimum-cell proof.
Android, sealed-package, final integrated review, and release gates remain pending.
No CI, push, tag, or publication started for this RN candidate task.

The RN candidate branch now contains the reviewed platform and workflow compiler changes.
Current clean commit: 67f95489a2b9a5b6d564cebbd36f8299248a352c.

Change

Current Swift Candidate selects Xcode 27.0, build 27A266a.
Release package and public consumer checks bind both declared RN tuples.
Current Android Candidate guards require the selected guest to report SDK37 and REL.
Each Android guard and its complete Make gate share one explicit Bash command.
Source quality starts with pinned actionlint and standard workflow discovery.
RELEASE.md:224 states the Xcode27 simulator limit and the separate iOS16.4 native gate.
The stale RN0.83-only test-context comment was removed.

Astra accepted the complete final platform and compiler source proposals.
The primary inspected and integrated their complete diffs.

Local executable evidence

make lint-workflows passed on clean 67f95489a2b9a5b6d564cebbd36f8299248a352c.
Initial and final source identities matched.

The exact pinned action parser produced one unchanged command for each folded Android input.
The parser came from commit a421e43855164a8197daf9d8d40fe71c6996bb0d with verified Git blob identity.

All 14 shell controls passed on that clean source.
Incorrect properties, transport failures with matching stdout, and a filter failure prevented Make.
Matching probes reached the unchanged Make command.
The simulated Make failure remained exit42.
These isolated shell controls use diagnostic commands and provide no native test result.

Both committed compiler negative controls failed as required:

Invalid expression in a .yaml workflow: clean f0ae3fa790f40f2f4c5d698338ce5e231be69b94, Make exit2.
Undeclared runner label: clean cb43fa9b6e8e671c836800f355853e0a6f14939b, Make exit2.

The exact xcode-27 allowance passed without changing hosted workflow routing.
Full integrated Mac source-quality execution remains active.
The desktop currently fails SSH banner exchange.
The complete Android and iOS16 minimum gates remain pending.
This task has not started CI or published the candidate.

Both complete local source-quality gates passed on clean 67f95489a2b9a5b6d564cebbd36f8299248a352c.
The primary inspected retained structured results and exact initial/final source identities.

Host	Go test events	Python executions	Rust tests	RN Jest tests	Kotlin JUnit tests
This Mac, Xcode27.0 build27A266a	2166	127	78	239	718
Desktop10.0.0.86	2167	127	78	239	718

Every required test executed.
No test failed, skipped, or produced an unexplained flaky outcome.
Both commands returned exit0.
Initial and final source identities matched, and both worktrees remained clean.
Rust executed 78 core tests and an empty auxiliary documentation harness.
The strict Rust parser accepted the positive aggregate and zero failure, ignore, and filter counts.
The first local summary observer incorrectly required a positive count in the empty documentation harness.
Its failed assertion remains recorded and was corrected without changing any gate or test.

The Mac run completed all 27 source-quality targets in their declared order.
The desktop used a separate committed checkout while the broad mutation run retained its original source.
The documented test prerequisites, lockfiles, checks, and parser controls remain unchanged.

Retained summaries:

.ignore/release-rc2/mac-source-quality-67f95489-summary.json.
.ignore/release-rc2/desktop-source-quality-67f95489-summary.json.
.ignore/release-rc2/dispatch/mac-final-source-quality-report.md.

The complete RN iOS Candidate gates are now in preparation on the owned iOS27 simulator with Xcode27.
The current Android and iOS16.4 minimum runtime gates remain pending.
Source-quality success does not replace those native or package gates.
This task has not started CI or published the candidate.

Complete Mac source quality passed on clean commit 2c8bd17c0ad53031dae1b4a28adbbc8f0bd8587e.
Xcode remained 27.0, build 27A266a.

Validation

Command: make ci-source-quality PGRX_TARGET_DIR=/private/tmp/synchro-rc2-pgrx-20261004.
The complete target returned 0.
Its initial and final source hashes matched, and both tracked statuses were clean.
The primary independently parsed the retained results:

Go: 2175 test executions passed, with zero failures or skips.
Python: 127 test executions across 12 harnesses passed.
Rust core: 78 tests passed, with zero failures, ignored tests, or filtered tests.
React Native Jest: 239 tests across 10 suites passed, with zero pending tests, failures, or runtime errors.
Kotlin: 718 tests across 46 JUnit files passed, with zero failures, errors, or skips.

Fresh adapter, seed, and extension artifact preparation also returned 0 on this exact clean commit.
The previous 67f95489 artifact identities remain unchanged.
Raw records use mac-integrated-2c8bd17c-20261004T224429Z under the owned dispatch directory.
Structured client results remain in .ignore/release-rc2/evidence/mac-integrated-2c8bd17c-20261004T224429Z-source-quality.

Retained native attempt

The earlier complete RN 0.82.1 attempt on 67f95489 passed fixture preparation, lint, and 239 unit tests.
It failed before native compilation because the local runner created a 162-byte Watchman socket path.
Its full logs, exit 2, current tuple evidence, and successful owned-service cleanup remain retained.
The next runner will use a short owned temporary directory.
Watchman, assertions, test selections, and SDKs remain unchanged.

Both complete native RN tuple gates and the separate iOS 16.4 minimum gate remain required.
Integrated desktop validation is in progress.
No CI, tag, pull request, or publication has started from this RN preparation task.
Issue #303 remains open.

The RN 0.82.1 app builds with fmt 12.1.0 under Xcode 27.
The complete candidate gate then fails its iOS smoke suite.
This result does not certify the minimum runtime or permit CI or release.

Execution

Source: clean 2c8bd17c0ad53031dae1b4a28adbbc8f0bd8587e before and after execution.
Host: this Mac, Xcode 27.0 build 27A266a.
Owned simulator: iOS 27.0, 0B03ABF6-AC85-46BE-9A69-BE241BEC19BE.
Command: complete make ci-candidate-rn-ios RN_RUNTIME_VERSION=0.82.1.
The runner uses the verified fresh server artifacts and a fresh PostgreSQL fixture.
The short owned temporary path resolves the retained Watchman setup failure.
Fixture preparation, lint, 239 unit tests, and native app build pass.
The native lock resolves React Native 0.82.1 and fmt 12.1.0, including the Folly requirements.
Smoke result: 27 executed, 17 passed, ten failed, zero pending or skipped, across two suites.
The smoke target returns exit 1. The complete candidate target returns exit 2.
Owned adapter and PostgreSQL cleanup both return exit 0.

Observed failures

Eight sync-flow tests fail.
Several expose NotStartedError: Sync has not been started through the native bridge.
The lifecycle failure occurs after the harness records lifecycle:foreground.
Push/pull, conflict, atomic-write, and offline-first failures also expose that error.
The busy-main-run-loop messages are observations, not an established root cause.

Two conformance host tests fail the conformance-harness visibility assertion.
Detox reports that the view does not meet its 75 percent visibility threshold.
The worker is inspecting the completed raw results and relevant source before selecting a correction.
No synchronization setting, assertion, production behavior, or test selection has changed.
No replacement attempt has started.

The subsequent bridge, package consumer, predecessor upgrade, scenario, and final runtime gates did not execute.
The RN 0.83.10 candidate gate did not start after the failed minimum gate.
The iOS 16.4 minimum gate also remains pending.

Retained execution prefix: mac-rn-ios-integrated-2c8bd17c-082-20261004T232830Z.
Raw Jest result: .ignore/rn-runtime/0.82.1/clients/react-native/example/artifacts/ios-test-results.json.
The complete log, result copies, source records, and prior failed attempts remain preserved.

The failed Mac attempt used incorrect local endpoint and authentication bindings.
This finding changes the diagnosis of that attempt. Its failed results remain retained.

Exact clean source: 2c8bd17c0ad53031dae1b4a28adbbc8f0bd8587e.

clients/react-native/example/src/App.tsx:26 selects port 8091 for the iOS example.
App.tsx:31 contains the documented example test token.
The retained Candidate command selected SYNCHROD_PG_PORT=9182 and SYNCHRO_TEST_PORT=9182.
The local runner also used the private conformance JWT secret for the additional example adapter.
A private HMAC check confirmed that the example token matches the declared default test secret.
The same check rejected the runner's private secret. No secret or token value was printed.

These mismatches establish a local execution setup defect.
The initial start case does not prove readiness. It only awaits lifecycle arming before stop.
Native start can return while its durable retry remains pending.
The later step labels also do not identify which native call failed.
Multi-user and seed-resume failures do not establish measured NOT_STARTED outcomes.

The primary stopped the proposed debugger run before any Make reproduction.
Complete native validation now uses the unchanged example's actual port and test-authentication bindings.
Private conformance credentials remain unchanged for the separate provisioned adapter.
No production code, token, endpoint constant, assertion, or native retry behavior changed.

The previous 27-test result remains failed, with 17 passes, ten failures, and zero skips.
This setup correction does not prove that all ten failures have one cause.
The two conformance launch-visibility failures remain separate outcomes.
Source-only lifecycle interleavings remain unmeasured and do not justify a speculative production correction.

Full native gates and release acceptance remain pending. No CI or publication has started.

The requested minimum-version consumer build passed on clean 6e89fe57e512aba997b708f4869e07af423fca55.
Command: make test-consumer-rn-ios SYNCHRO_RN_VERSION=0.82.1, within the full RN iOS Candidate.
The owned helper verified the exact clean source before starting the Candidate.
The source remains clean and unchanged while that complete gate runs.

The consumer installed React-Core 0.82.1 and fmt 12.1.0.
Xcode 27 completed the device and simulator builds.
The consumer target reported Packaged React Native iOS consumer build passed.
The RN 0.83.10 consumer target also passed.
The local log preserves the dependency resolution and both complete build results.

This result validates the fmt backport for the RN 0.82.1 consumer build.
It does not replace runtime, upgrade, corpus, packaged-cell, security, or Candidate checks.
The full RN iOS Candidate has advanced to its package upgrade gate.
The RN Android corpus repair remains in diagnosis.
No PR, CI, merge, tag, release, or publication has occurred.

Validation on e2c3024380e288a65efdf2f9cbae27b7aefe87c0

make ci-source-quality passed in 931 seconds on desktop 10.0.0.86.
The worktree remained clean before and after the gate.
The gate includes documentation, version, package, build, lint, contract, harness, and unit checks.

Kotlin: 748 tests in 46 JUnit suites. Zero failures, errors, or skips.
React Native: 269 unit tests. Zero failures or pending tests.
Fresh Trivy scan: exit 0 with zero SARIF findings under the unchanged CI policy.

Structured Kotlin and React Native results were retained in the owned result directory.
Unrelated installed SQL remained unchanged.
Only the verified quality-owned ADB service was stopped. It had no connected devices.

Full server, Swift, Android, React Native, and packaged support-cell gates remain active or pending.
The known unfixed dependency advisories remain open. This result does not waive them.
No release certification is claimed.

Full Swift validation on e2c3024380e288a65efdf2f9cbae27b7aefe87c0

make ci-candidate-swift passed in 1,247 seconds on the local Mac with Xcode 27.
The complete helper exited 0 after fixture cleanup and clean-worktree checks.

Complete Swift semantic corpus: 15 tests passed.
Native xcresult: 393 tests passed.
Shared schema identity: one Swift test and one Kotlin test passed.
Isolated packaged Swift consumer: build passed.
Published predecessor to candidate upgrade: exact target passed.
Structured parsers reported no failures or skips.

Test results and tested products were copied and hash-verified before completed cache removal.
The main Swift archive retains 901 files.
Each upgrade archive retains 17 product files.
The Mac had 4.7 GiB free after cleanup.

Full React Native iOS and Apple support-cell validation has started on the same clean commit.
Server and Android validation remain active.
No push, tag, release, or master merge has occurred.

Full Kotlin validation on e2c3024380e288a65efdf2f9cbae27b7aefe87c0

make ci-candidate-kotlin passed on desktop 10.0.0.86.
The helper confirmed a clean worktree after the phase.

Device instrumentation: 8 tests passed.
Complete native semantic corpus: 16 tests passed.
Extension-backed integration: 44 tests passed in four JUnit suites.
Packaged consumer: one device test passed and the isolated consumer build passed.
Published predecessor to candidate upgrade: exact target passed.
Structured parsers reported no failures or skips.

The retained native XML reports contain 52 tests with zero failures, errors, or skips.
The first report-copy attempt used the wrong instrumentation directory and failed.
The corrected collector used the Makefile's conformance-app result path and passed.
No test, selector, assertion, or gate changed.

Full React Native Android validation is now active on the same clean commit.
Full React Native iOS, server, and packaged support-cell validation remain active or pending.
No complete release certification is claimed.

Validation update

The complete server run on e2c3024380e288a65efdf2f9cbae27b7aefe87c0 failed after 4,090 seconds.
Its binding diagnostic and readiness throughput failures remain recorded.
The run did not reach the seven integration mutants or the server support cell.

Commit 75de88976420cdd9d5b59ea8915bb80aa7b88ad5 preserves the last native resolution error before cancellation.
The focused real binding proof passed in 87 seconds on that clean commit.
Both corrupt captures failed image verification. The restored capture passed.
This focused run used PARTIAL=1. It does not certify the complete gate.

Commit cc4ae3c26b5156342984bc2216074aa9683435b8 passed 254 harness tests and 1,252 driver tests.
The focused readiness run passed in 41 seconds on that clean commit.
All 240 writes completed inside the unchanged 30-second window.
Maximum scheduled-start delay was 207.620 milliseconds.
Total successful execution duration was 1.974 seconds. Maximum execution duration was 239.163 milliseconds.
All 150 readiness samples passed.
These measurements do not establish the cause of the earlier throughput failure.

The independent review found a timing log outside the existing assertion boundary.
Commit b536bbf94a125d58629a7532975342c93b733a5b moves that output into existing diagnostics.
The correction does not change the workload, threshold, SQL, assertion, or mutation classifier.
Independent review and complete validation of this correction remain pending.
The early-publication negative control is running separately. Its deliberate defect will not enter the work branch.

Android update

The full Kotlin and React Native Android candidate phases passed on clean e2c3024380e288a65efdf2f9cbae27b7aefe87c0.
The React Native phase passed 27 smoke tests, 16 bridge tests, both consumer version builds, upgrade validation, and the complete corpus.
The later packaged-fixture preparation failed because its registry did not activate.
The overall helper result remains failure, with successful cleanup and unchanged source.

A separate owned fixture now runs the four Android support cells with the verified candidate artifacts.
Fresh fixture preparation passed. This matches the separate fixture boundary used by CI.
No registry activation requirement was removed or weakened.

Remaining work

Complete the fresh server gate and its mutation controls.
Complete both platform cell runs and the active iOS candidate run.
Complete source quality and dependency scanning for the final source commit.
Preserve current dev commits before the work branch push and pull request.
Keep unresolved dependency advisories visible. No security exception is approved.

Archive reuse correction

The first independent review rejected 24b72d55db42a66d1897cf5b51134ecf9a1c00d5.
Persistent cached bytes could change between hashing and extraction.
The rejected review remains in the validation records.

Commit cfdd99f7e3037c3efc7a0e6dfab7815d3bf8e666 restores the private-directory boundary.
It copies cached bytes into the existing private archive before size and SHA-256 verification.
Verification and extraction use that snapshot.
The correction adds no configuration option, dependency, helper, fallback, or validator exemption.

Independent review accepted the correction and its deletion check.
Primary validation on the exact clean commit passed make test-android-emulator-prepare.
The structured result contains 89 tests, zero failures, zero errors, and zero skips.
The prerequisite result also passed.

Demonstrated negative control

The deliberate mutant 102d2075a1bd7e0170ce6c22fb0f538baf29f078 restores direct extraction from the persistent cache.
The same complete Make target ran 89 tests and failed exactly one test, with zero skips.
test_cache_replacement_after_hashing_installs_private_snapshot observed the replacement executable instead of the original fixture executable.
This proves that the control detects the rejected behavior.
The mutant remains outside the work branch and will not be pushed.

Runtime Android cell validation remains pending on the corrected commit.
The earlier download failure and successful earlier cells keep their original source identities.
Source-quality validation on clean cc4ae3c26b5156342984bc2216074aa9683435b8 passed in 901 seconds.
Its Kotlin results contain 748 tests in 46 suites. Its RN unit result contains 269 tests.
All structured results have zero failures, errors, and skips.
Final source-quality validation remains required after the runtime cells finish.

Complete RN iOS candidate result

make ci-candidate-rn-ios passed on clean e2c3024380e288a65efdf2f9cbae27b7aefe87c0 in 7,784 seconds.
The host used Xcode 27 and the installed iOS 27 simulator.

RN smoke: 27 passing tests, zero failures or skips.
Native bridge: 11 passing tests, zero failures or skips.
RN 0.82.1: device and universal simulator consumer builds passed with fmt 12.1.0.
RN 0.83.10: device and universal simulator consumer builds passed.
Native package upgrade: the exact required target passed.
Complete RN iOS corpus: 27 named results passed in one package, with zero failures or skips.

The source remained clean. Candidate artifacts then built successfully in 11 seconds.
The subsequent packaged-fixture restart failed in 94 seconds because its client registry did not activate.
The overall helper result remains failure status 2. Its candidate result remains a separate pass.
This failed setup does not establish a passing lifecycle cell.

Fresh cell fixture

A separate owned fixture now runs the four Apple lifecycle cells with the completed artifacts.
Its extension manifest, adapter and seed sidecars, copied bytes, and client artifact hashes were verified.
Fresh fixture preparation passed. The iOS 17 cell is running.
The registry activation requirement remains unchanged.

Completed RN upgrade and main iOS build caches were removed after their gates passed.
Both upgrade applications, the final corpus application, debug symbols, bridge results, and hashes remain preserved.
Mac free space increased to 12 GiB before the Apple cell builds.
Active build files and unrelated user files were preserved.

Final source, server, and cell results remain required before the branch push and CI.

Android lifecycle cells passed

All four required Android cells passed on clean cfdd99f7e3037c3efc7a0e6dfab7815d3bf8e666.
The run used the audited private-snapshot archive correction.
It preserved the previously built native artifacts and verified their hashes.
Native client and package source inputs are unchanged from producer commit e2c3024380e288a65efdf2f9cbae27b7aefe87c0.

SUP-ANDROID-MIN-001: Android API 24.
SUP-ANDROID-CURRENT-001: Android API 37.
SUP-RN-ANDROID-MIN-001: RN 0.82.1 on Android API 37.
SUP-RN-ANDROID-CURRENT-001: RN 0.83.10 on Android API 37.

Every cell records passed terminal connect, push, pull, kill, and resume operations.
Each operation has a positive test count.
Every cell records two durable pending writes before termination and zero after resume.
Every cell verifies the remote value and the resumed write on the real server.
Initial and resumed process identifiers differ.
The pinned emulator is version 37.2.12, build 16428233.

The helper returned zero. Cleanup passed, source stayed clean, and unrelated SQL stayed unchanged.
These are local source-consumer results. They are not sealed release-package results.
The earlier failed downloads and setup attempts remain visible with their original source identities.

Dependency scan

The fresh configured Trivy scan passed on the same clean commit.
Its exit status is zero, and its SARIF contains zero findings.
The CI scanner flags and advisory policy remain unchanged.
This result does not grant an exception for the separate unresolved dependency advisories.

Final source quality, Apple RN cells, complete server validation, and the readiness negative control remain pending.

Final source quality and audits

make ci-source-quality passed on clean cfdd99f7e3037c3efc7a0e6dfab7815d3bf8e666 in 461 seconds.
Its preserved Kotlin results contain 748 tests in 46 suites.
Its RN unit results contain 269 tests.
The structured results contain zero failures, errors, and skips.
The completed owned ADB service had zero connected devices and stopped successfully.

The configured Trivy scan passed on the same clean commit with zero SARIF findings.
Its policy is unchanged.

Both full package-manager audits returned status 1 on that clean commit.
The RN audit still reports High Braces and Moderate sprintf-js and stream-json advisories.
Deprecation reports also remain visible.
The documentation audit reports ten Moderate affected records through postcss-selector-parser.
The named compression, Handlebars, and shell-quote fixable advisories are absent.

Current upstream evidence

Official npm metadata reports Braces 3.0.3 and sprintf-js 1.1.3 as their latest versions.
Those versions remain inside their reported affected ranges.

Stream-json 3.7.0 exists, but the declared consumer ranges exclude it.
Detox 20.51.4 still declares stream-json ^1.7.4.
The authored lock also contains bunyan-debug-stream 3.1.1 with stream-json ^1.7.5.
The latest published stream-json 1.x version remains 1.9.1.
The full audit reports depth, JSONC comment scanning, and Assembler prototype-pollution advisories for that version.
No forced major replacement or advisory suppression was added.

Expressive Code core 0.44.2 still declares postcss-nested ^6.0.1.
Postcss-nested 6.2.0 declares postcss-selector-parser ^6.1.1.
The separate documentation finding remains open.

The raw audit failures and registry metadata remain preserved.
No security exception or publication approval is granted.
Related blockers: #312, #328, #330, and #319.

Mac support cells passed

The four local Mac cells passed on clean e2c3024380e288a65efdf2f9cbae27b7aefe87c0.
The completed helper returned status 0 and cleaned up its owned adapter and PostgreSQL cluster.

SUP-IOS-MIN-001: iOS 17.0 with Xcode 27.0.
SUP-IOS-CURRENT-001: iOS 27.0 with Xcode 27.0.
SUP-RN-IOS-MIN-001: RN 0.82.1 with iOS 27.0 and Xcode 27.0.
SUP-RN-IOS-CURRENT-001: RN 0.83.10 with iOS 27.0 and Xcode 27.0.

Each structured cell record reports successful connect, push, pull, process termination, and resume operations.
Each operation executed one test.
Each process restart used a different process identifier.
Each durable queue changed from two pending records before termination to zero after resume.
Each cell verified the applied remote value and the resumed server write.

The raw cell records, measured environments, artifact hashes, and cleanup results remain preserved.
These results certify local source consumers. They do not certify sealed release packages.

The SDK and package inputs are unchanged between that commit and cfdd99f7e3037c3efc7a0e6dfab7815d3bf8e666.
git diff --exit-code e2c30243..cfdd99f7 -- clients Synchro.podspec Package.swift returned status 0.
The later commits change server proof diagnostics and emulator archive handling only.

The earlier candidate helper restart failure remains visible in the previous comment.
This successful run used the separate fresh fixture required by the support-cell procedure.
No registry activation check was weakened.

Master, tags, and published packages remain unchanged.

Complete local server validation passed

make ci-candidate-server ADAPTER_TEST_URL=postgres://127.0.0.1:1/not-owned passed in 5083 seconds.
make test-consumer-server passed in 19 seconds.
Both ran on clean b536bbf94a125d58629a7532975342c93b733a5b on desktop 10.0.0.86.
The complete helper returned status 0.
The unrelated installed SQL checksum remained unchanged.

The candidate records 435 Rust tests across three harnesses.
Its normal Go JSON records contain 1319 passing test events, zero failure events, and zero skip events.
The integration mutation gate reports seven detected mutants and zero surviving mutants:

cursor-advancement
wal-acknowledgment
mutation-conservation
checksum-correctness
scope-isolation
progress-order
pull-deduplication

The readiness load completed 240 commits with 150 samples.
Maximum heartbeat age was 0.735 seconds.
Maximum WAL lag was 1237360 bytes.
Maximum commit lag was 0.632 seconds.
Maximum scheduled start delay was 1.170197 milliseconds.
The original workload, sample interval, assertion, and acceptance thresholds remained unchanged.

SUP-PG-LINUX-X64-001 passed on Ubuntu 24.04 x86_64 with PostgreSQL 18.3.
Its structured record verifies connect, push, pull, process termination, and resume.
Each operation executed one test.
The restarted process used a different identifier.
The record verifies equal replay, the applied remote value, and the resumed server write.

Server and conformance inputs are unchanged from this commit to cfdd99f7e3037c3efc7a0e6dfab7815d3bf8e666.
git diff --exit-code b536bbf9..cfdd99f7 -- conformance extensions api returned status 0.
The later delta changes emulator archive handling and its documentation only.

The separate focused readiness mutant remains active.
This comment does not claim its result.
Earlier failures remain preserved in the issue history.
Master and publication remain unchanged.

Focused readiness negative control passed

make test-integration-mutant INTEGRATION_MUTANT_ID=readiness-position-equality passed in 244 seconds.
It ran on clean b536bbf94a125d58629a7532975342c93b733a5b on desktop 10.0.0.86.
The owned helper returned status 0.
It used the completed server gate artifacts from that same commit.

The gate reports:

KILLED readiness-position-equality by TestRealReadinessStaysReadyUnderWriteLoad/assertion
Focused integration mutation control passed: readiness-position-equality

The structured mutant result failed the named assertion.
The failure was not an environment or setup failure.
The normal baselines passed before and after the mutant.
This result verifies the corrected assertion boundary.
It does not claim execution of the complete extended mutation manifest.

The full normal server gate and seven required integration mutants already passed on the same commit.
Its server and conformance inputs are unchanged at final source cfdd99f7e3037c3efc7a0e6dfab7815d3bf8e666.

The PR evidence review accepted the corrected claims and completed its deletion and simplification check.
The local Mac and desktop checks are complete.
The upstream dependency findings remain open and block publication.
Master, release tags, and published packages remain unchanged.

Merged into dev

PR #333 merged as 538e1bf7d302d098d33659cb507a2489de761159.
It preserves prior dev cd28ae93c9604f82bbc92ba5cd3cfe034e5dc648 and source cfdd99f7e3037c3efc7a0e6dfab7815d3bf8e666.
All approved RN, fmt, and native repairs remain included.

CI run 38029034667 completed successfully before the merge.
CI checked synthetic merge 6634927febe915ed40e7f6024279d14cb9a82c47.
Its source tree equals both the tested source and the actual dev merge tree:
1294b9f7a9fbdbb0900d03746275bb2ad3f92811.

Source quality passed, including 269 RN tests, 748 Kotlin tests, and 89 emulator tests.
Swift unit validation passed with 363 tests in the structured xcresult.
The configured dependency scan passed with its unchanged policy.
Go, JavaScript, Kotlin, and Swift CodeQL analyses each report zero findings and an empty analysis error.
The required pull-request aggregate passed.

The full candidate jobs are restricted to master pushes by the authored workflow.
They did not run for this PR.
The complete local candidate and support-cell results remain tied to their exact source commits in earlier comments.
This comment does not replace those results with a claim of complete candidate CI.

Master remains 4f7967b3b23dd7da4dab226aa2a09baf89c0a219.
PR #317 remains a draft, and its promotion checks rerun automatically for the new dev head.
The remaining upstream dependency issues still block publication: #312, #328, #330, and #319.
No master merge, release tag, publication, security exception, or protected approval occurred.

Fixed in 538e1bf7d302d098d33659cb507a2489de761159

Change

The supported RN range is 0.82.1 through 0.83.x.
RN 0.82.1 iOS consumers use the documented upstream fmt 12.1.0 backport with Xcode 27.
Peer ranges, support cells, native package requirements, workflows, and documentation agree.
The approved Apple minimum is iOS 17.
Unmodified RN 0.82.1 is not represented as passing this Apple toolchain.

Source: clients/react-native/package.json:88, clients/react-native/.yarn/patches/react-native-npm-0.82.1-c9dce9d96d.patch, conformance/support-matrix.json:133, and RELEASE.md:233.

Validation

Complete make ci-candidate-kotlin passed on clean dd2d77f1491cfea88b31613b47998ecc2b4734c9.
Complete make ci-candidate-rn-android passed on clean e2c3024380e288a65efdf2f9cbae27b7aefe87c0.
Complete make ci-candidate-rn-ios passed on that clean source with Xcode 27 and iOS 27.
The iOS Candidate passed 27 smoke tests, 11 native bridge tests, and 27 corpus results, with zero failures or skips.
RN 0.82.1 device and universal simulator consumer builds passed with the exact fmt backport.
RN 0.83.10 device and universal simulator consumer builds also passed.
All four Apple local lifecycle cells passed on clean e2c3024380e288a65efdf2f9cbae27b7aefe87c0.
All four Android local lifecycle cells passed on clean cfdd99f7e3037c3efc7a0e6dfab7815d3bf8e666.
The cells cover iOS 17, current iOS 27, Android API 24, current API 37, and both RN versions.
Each cell passed connect, push, pull, process termination, and resume through the real extension-backed adapter.
git diff --exit-code e2c30243..538e1bf7 -- clients Synchro.podspec Package.swift returned zero.
git diff --exit-code cfdd99f7..538e1bf7 returned zero.
make version-check, make check-released-update-scripts, and the merged PR checks passed on the clean integrated source.
Independent closure review accepted the literal criteria and completed a separate simplification pass.

Execution evidence: complete RN iOS Candidate, Apple lifecycle cells, Android lifecycle cells, and integrated CI evidence.

The original fmt compiler failure and every later failed execution remain preserved.
This closes RN minimum adoption and its local certification criteria.
It does not close the separate dependency findings or replace the required sealed Package gate.

Navigation Menu

Repository navigation

RN 0.82.1 certification requires a fmt 12.1.0 backport #303

Finding

Initial source and evaluation scope

Initial executable evidence

Upstream evidence

Required adoption changes

Updated recommendation and remaining work

Acceptance criteria

RN 0.82.1 application evaluation

Complete application builds

Reproducible patch and unit checks

Native test harness failure

Native execution results

Remaining certification scope

RC.2 local validation update

Completed evidence

Retained failures and active work

35 remaining items

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

RN 0.82.1 certification requires a fmt 12.1.0 backport #303

Description

Finding

Initial source and evaluation scope

Initial executable evidence

Upstream evidence

Required adoption changes

Updated recommendation and remaining work

Acceptance criteria

Activity

mdspinali commented on Oct 4, 2026

RN 0.82.1 application evaluation

Complete application builds

Reproducible patch and unit checks

Native test harness failure

Native execution results

Remaining certification scope

mdspinali commented on Oct 4, 2026

mdspinali commented on Oct 4, 2026

mdspinali commented on Oct 4, 2026

mdspinali commented on Oct 4, 2026

RC.2 local validation update

Completed evidence

Retained failures and active work

mdspinali commented on Oct 4, 2026

mdspinali commented on Oct 4, 2026

mdspinali commented on Oct 4, 2026

mdspinali commented on Oct 4, 2026

mdspinali commented on Oct 4, 2026

mdspinali commented on Oct 5, 2026

mdspinali commented on Oct 5, 2026

35 remaining items

mdspinali commented on Oct 9, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 10, 2026

mdspinali commented on Oct 11, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions