Repository navigation
Conversation
|
One parallel diagnostic validation is selected on frozen source Android asks which actual startup boundary is reached before the missing passcode screen. iOS asks whether Intel compilation completes within the reviewed compile-only budget and then passes app startup/product checks. The source stays fixed through this run; no additional identical-source retry is selected or authorized. No personal Mac is required. Earlier5bb failures remain: Android built/installed but failed Create-passcode; iOS driver passed but compilation exited124 at2400.3seconds. The exact5bb software624/624 proof remains attributed to5bb. No result transfers to this head. Both producers and the final18-path integration are independently reviewed. All462 Python controls,851 release-origin controls,36 actual pinned-SDK controlled cases and12 draft-policy controls pass as scoped in the PR. The local app command's11 unchanged Ruby/config prerequisite failures remain recorded; actual configured Tests verification is pending. Production profiles/windows remain empty and the known High still blocks operation. No installed canonical, physical, signing, upload or public-release acceptance is claimed. |
Both apps now build on hosted cloud runners. On the preceding exact source
2b6c3ce56857b5dc92f17cc37453ff308d5a3ae8, Android also passed startup, native persistence and scoped R5 worker probes, but its first-wallet sample stopped after onboarding Skip. iOS compiled in 42m56s and installed successfully, then the OS refused launch. The installed iOS app matched the build and passed signature/entitlement checks; the launch cause remains unknown.This successor corrects a proven Android evidence-reader defect: released Maestro 2.11.0 logs three Kotlin-mangled method names, which the source-name-only parser rejected. Four cases executed against the actual released JVM classes reproduce baseline failure and corrected extraction, including nested/retried flows. The existing reader can now bind the selected Skip bounds and actual driver tap point. App behavior, taps, selectors, waits and output schema are unchanged.
For iOS, one failure-only direct simctl launch runs on the same owned, installed app after the original failure observations are frozen. It has a 20-second limit and exports only fixed OS domains/codes and owned process observations. Original Maestro failure remains authoritative even if this later launch succeeds. No later acceptance phase can be unlocked. If the next run still yields only generic FBS1/UNKNOWN, stop repeating this ephemeral CI diagnosis and prepare persistent cloud-Mac debugging.
The candidate also retains the reviewed canonical/private evidence collector and protected current-source operation bridge. A finite f0 producer profile is now prepared in source; live installation/windows and read identities remain separate. The duplicate-JSON security test fixture is constructed without first-occurrence replacement, and a missing observer dependency is supplied to an existing VM test. No app/native dependency, release counter or private R5.2 code changed in this successor.
Exact head:
47a92a4cdd9081d39c918a07a2d3a9d2607c9aaa; tree:03c8957dceb8088339cb8e61f3d3bf208848f171. All 21 changed paths against 2b6 match five independently reviewed producers exactly. Combined verification passes 472 Python tests, 63 affected Node contracts, the complete 2,518-file raw-source guard and diff checks. Earlier full 2,159 Node/42 pinned-SDK controlled cases belong to eb02; JavaScript/CJS/workflows are byte-identical from that reviewed assembly. Actual hosted Tests job121102810527 in run40171405919 succeeds on this source. The separate legacy matrix fails, so the overall CI run is FAILURE. CodeQL analysis job121102809817 succeeds; the official security check is NEUTRAL with no annotations and an unmatched base configuration after the category change. This is not full security clearance. Exact-source native execution remains in progress.Current parallel native run: https://github.com/KeeperCommunity/bitcoin-keeper/actions/runs/40171436126 (attempt1), Android job121102901366 and iOS job121102901409. Source is frozen; no identical retry is authorized.
Historical native run: https://github.com/KeeperCommunity/bitcoin-keeper/actions/runs/40166008563 (attempt1). Android job121087085987: build/install/startup/native-persistence/scoped-R5 PASS, first-wallet onboarding FAIL. iOS job121087085817: driver/build/install PASS, startup launch FAIL, later probes NOT RUN. These results belong to 2b6 and do not transfer native acceptance to this successor. Original failures and raw-artifact digest receipts are preserved.
The previous source's hosted Tests job121087003121 passed; the separate legacy-v2 backup matrix remained 2 PASS/293 FAIL, so overall CI failed. The previous official security annotation was js/incomplete-sanitization in the test fixture now corrected; official clearance of this successor and inherited security disposition remain unclaimed. Diagnostic native results cannot provide protected-base acceptance until reviewed infrastructure/source adoption. Physical acceptance remains NOT_RUN_NOT_WAIVED; signing/store access and exact owner operation approval remain required.
Superseded testing PRs #7149, #7154, #7157, #7158, #7159, #7162 and #7163 remain closed as superseded without acceptance. Their branches and original evidence are preserved. No merge into sprint, live profile/window installation, release signing, store upload, distribution or deployment occurred.
Codex local handoff
State: NEEDS_LOCAL
PR: #7164
Base: sprint
Head branch: codex/r5-cloud-release-evidence-20261025
Head commit: 47a92a4
Goal and acceptance criteria: complete Android first-wallet onboarding and iOS app testing in the cloud, then authenticate required installed native/security evidence before any separately approved artifact operation.
Changes made: exact released-Maestro tap log aliases, bounded failure-only iOS launch observation, finite private producer profile and two reviewed CI fixture corrections.
Cloud checks: current combined source checks pass as scoped above; hosted Tests and CodeQL analysis succeed, overall legacy-matrix CI fails, security comparison is inconclusive, and the selected exact-source native jobs remain in progress. Previous 2b6 native outcomes remain separately attributed.
Required local checks: IOS, ANDROID, HARDWARE
Local checks requested: exact-source native/public-private acceptance on appropriate cloud runners and separately scoped physical checks. No personal Mac is required.
Risks / known gaps: unresolved app/UI outcomes, protected source/infrastructure adoption, live private windows/read identities, security disposition, release access and physical acceptance.