Skip to content

AA Experiments v1.0.0-beta4

Latest

Choose a tag to compare

@github-actions github-actions released this 27 Oct 16:36
· 1 commit to main since this release
c80b857

AA Experiments 1.0.0-beta4

Application ID: io.github.highwindbr.aaxp. Installed version: 1.0.0-beta4 (versionCode=169).
Official release APK: AAExperiments-1.0.0-beta4.apk with a corresponding SHA-256 checksum.

This beta improves reliability and ergonomics of Android Auto experiment
inspection, configuration and recovery. It does not unlock every
Android Auto, server or vehicle capability.

What's new

  • Advanced Overrides: opt-in editing of a strictly eligible,
    structurally verified subset of BOOLEAN, LONG, DOUBLE and STRING
    getters. Explicit selection, two-stage review, independent saved
    baselines, Pause / Resume / Restore and risk warnings are provided.
    Advanced entries use the current Experiment details UI.
  • Stronger compatibility: the frozen Android Auto 17.8 fixture
    preserves all 919 beta3-editable keys and adds 23 individually
    reviewed keys: 12 parked-mode media-timing, three WirelessLatency
    false SSL matches and eight BOOLEAN config/manufacturer flags
    previously misclassified as possible Protobuf. All still require
    valid getter evidence; risky parked-mode timing values require
    the exact PARKED APPLY confirmation, including in the normal
    Experiment editor.
  • Typed override safeguards: stricter normal-path decoding,
    correct DOUBLE compiled defaults, integer range checks and
    redaction of sensitive STRING returns in LSPosed logs. A fixed
    Advanced BOOLEAN selection bug now enables Review when True/False
    is chosen, without bypassing elevated-risk confirmation.
  • Runtime stability: two-slot remote profile publication,
    serialized refreshes, profile migration and fail-open recovery,
    pinned to the precise installed base/split Android Auto fingerprint.
  • Read-only diagnostics: reviewed getter probes and Video Timing
    Controller tracing under collapsible developer tools.
  • Clearer UI: theme-aware surfaces, adaptive Experiments layout,
    clearer configured/runtime terms, input validation and feedback.
    The compact analysis dialog uses bounded current-stage progress
    without reserved blank rows or detached end dots, and avoids
    transient Not found messages while scanning. Old catalog
    safety-policy descriptions are marked historical.
  • Backup & recovery: one Settings page prioritizes full recovery
    export/import, with legacy normal-only operations and the destructive
    reset in separate collapsed sections. The complete recovery bundle
    includes normal, Advanced and Experimental states and baseline
    absence; exact-build/type/ownership validation and typed
    confirmation protect restoration. Existing normal-only backup
    formats remain supported.
  • Large JSON/CSV exports: stream SQLite results in 250-row
    pages instead of buffering 50,000+ records in the Compose UI.
    Backup imports have bounded file sizes.

Validation and evidence

  • The signed 1.0.0-beta4-preflight22 candidate passed public
    CI 40325455245
    and private gates 40431677743:
    all three 256 MiB fixture regressions, non-skipped frozen 17.8
    full corpus, APK v2 signature and certificate match against the
    published beta3 APK. Later delivery-tooling and UI-acceptance
    changes passed CI 40524090180.
  • Preflight22 Android 17 / Galaxy S24+ screenshots established scan,
    Backup & recovery, Advanced BOOLEAN selection/confirmation and
    pause/restore UI behavior, not runtime getter-return evidence.
  • Earlier signed preflight21 LSPosed logs separately established
    normal, Advanced and Experimental getter interception and typed
    returns on Android Auto 18.0. The preflight22 visual fixes did not
    modify the underlying runtime hook algorithms.
  • The maintainer reported successful normal backup restoration
    and complete recovery restoration on 2026-10-26, then accepted
    preserved override state after the preflight22 update. No extra
    Status screenshots or export are required. Isolated disposable
    storage and exact byte-for-byte restored state were not
    independently verified
    .
  • Additional test evidence includes a 50,675-item full-catalog
    JSON/CSV comparison and a 20,147-byte recovery-bundle export.
  • The final 1.0.0-beta4 / 169 candidate passed public CI 40528898122
    and private signed gates 40532709266,
    including full-corpus and fixture regression, v2 signature, package
    metadata, and matching beta3 signing certificate. The maintainer
    installed this candidate and reported correct operation on
    2026-10-27. The official tagged-release workflow independently
    rechecks its own gates before publishing its built APK.

Compatibility and limitations

  • Advanced is not an unrestricted toggle. Unproven, unsupported,
    sensitive or incompatible mappings remain read-only/suspended.
    Getter interception alone cannot guarantee any head-unit, server
    or downstream feature will behave differently.
  • Complete backup files can contain raw STRING data; they are not
    encrypted or authenticated
    , must be stored privately and must not
    be imported from untrusted sources. Full restore requires the exact
    matching Android Auto build and should only be destructively tested
    in isolated/disposable storage.
  • Long scans can stall if the display turns off; keep the phone awake.
    LSPosed API 101 offers no Manager-side reverse telemetry, so check
    PROFILE_ACCEPTED, HOOK_INSTALLED and OVERRIDE_RETURNED in
    LSPosed logs.
  • Some CSV viewers mishandle raw control characters. Do not perform
    overrides or restart projection while driving.

AA Experiments is independent GPL-3.0-or-later software, not
affiliated with Google, Android Auto or LSPosed. See LICENSE,
THIRD_PARTY_NOTICES.md and docs/KNOWN_LIMITATIONS.md.