AA Experiments v1.0.0-beta4

This release introduces Advanced Overrides, improves compatibility with newer Android Auto versions, and strengthens runtime stability, diagnostics and recovery.

What's new

• Advanced Overrides: Safely configure eligible BOOLEAN, LONG, DOUBLE and STRING flags, with risk warnings, confirmation, independent baselines and Pause / Resume / Restore controls.
• Expanded compatibility: Preserves all 919 previously editable flags and adds 23 newly verified entries from Android Auto 17.8.
• Improved reliability: Safer typed overrides, stricter validation, atomic profile updates and automatic recovery from incompatible configurations.
• Better diagnostics: Read-only getter probes, Video Timing Controller tracing and improved experiment inspection.
• Refined UI: Adaptive layouts, theme-aware components, clearer status indicators and smoother catalog scanning.
• Backup & recovery: Complete export/import of normal, Advanced and Experimental configurations, with build validation and legacy backup support.
• Faster exports: Memory-efficient JSON/CSV exports for large catalogs.

Validation

• Passed public CI and private release gates, including regression tests, APK signature and signing-certificate verification.
• Tested on Galaxy S24+ (Android 17) with Android Auto 17.8 and 18.0.
• Verified override interception, backup restoration and large-catalog exports.

Important notes

• Advanced Overrides only supports verified mappings. Enabling a flag does not guarantee that the corresponding Android Auto feature will work.
• Keep the screen awake during long scans. Never modify overrides while driving.

See Known Limitations for further details.

AA Experiments v1.0.0-beta3

This beta improves custom-value feedback, fixed-height resolver progress, and suspended-override auditing while retaining the installed package ID and existing override keys.

Changes

• Category filters were returned to the Filters sheet and disappear when no entries exist.
• Added/Removed/Changed open actual editable-catalog build comparisons with previous/current details.
• Suspended overrides can be hidden from All until the Android Auto build changes, without deleting or activating them; they remain accessible through Suspended and the audit.
• Code packages and namespace now match io.github.highwindbr.aaxp.
• Feature Catalog updated to cover Android Auto 17.8, 17.9 and 18.0 daily.
• LONG, STRING and DOUBLE Apply feedback clears focus and briefly confirms a successful save, while errors retain the editable field.
• CUSTOM rows use a Lab Green chip and left-side indicator.
• The resolver dialog reserves three lines and breaks long identifiers at underscores.
• Suspended overrides are visibly distinct, searchable and removable after confirmation; flags are not force-reapplied.

Validation

Release build, unit tests, private fixture gates and a real-device smoke test must pass before publication.

Known limitations

The app may display a SHA-based identifier for older suspended overrides if no historical flag name can be recovered. Keep the device awake during long scans.

AA Experiments v1.0.0-beta2

Fixed

• The first APK import no longer reports a SQLite query error after successfully saving a catalog.
• Catalog retention now checkpoints the SQLite WAL through the query API before vacuuming.

Release readiness

• Adds regression coverage for pruning old catalogs while retaining the active catalog and its predecessor.
• Keeps existing runtime overrides and their preference storage unchanged.
• Retains API 101 entry point, static Android Auto scope and protective exception mode used by the first beta.

Known issue

Keep the device awake while a scan or deep analysis is running. Screen-off can still cause an
active analysis to stall or fail to complete. This accepted beta limitation does not invalidate
completed analysis results or runtime overrides generated from a completed profile.

Additional limitations and recovery guidance are documented in docs/KNOWN_LIMITATIONS.md.

AA Experiments v1.0.0-beta1

First public beta of AA Experiments, an Android Auto experiment inspector and runtime override
manager for LSPosed Modern API 101.

Highlights

• Installed Android Auto discovery and structural re-resolution without hardcoded obfuscated production names.
• Typed, safety-gated scalar overrides with incompatible or ambiguous mappings suspended before publication.
• Runtime override snapshots with live preference refresh and fail-open behavior.
• Composite installed-build fingerprinting across the base APK and splits.
• Release regression coverage against private Android Auto fixtures, including three isolated 256 MiB gates.
• Frozen Android Auto 17.8 full-corpus gate covering all 919 editable keys with zero blocking UNKNOWN callsites.
• Restore-all recovery path and explicit diagnostics for runtime profile acceptance and hooks.

Known issue

Keep the device awake while a scan or deep analysis is running. Screen-off can still cause an
active analysis to stall or fail to complete. This is a known, accepted beta limitation and does not
invalidate already completed analysis results or runtime overrides generated from a completed
profile.

Additional limitations and recovery guidance are documented in docs/KNOWN_LIMITATIONS.md.