v2.0.0-beta.3

What's Changed

If you were on gRPC and saw notify disconnects every ~90 seconds, power-on or exec commands failing after the Quad sat in standby for about an hour (often with preflight/INVALID_ARGUMENT in the logs while HA still looked connected), or entities staying “available” when the gRPC session had actually dropped, beta.3 targets those cases:

• keepalive pings no longer provoke those disconnects
• stale exec auth is refreshed automatically after long quiet notify
• entity availability should update when the session connects or drops.

⚠️ This transport method is still highly experimental and subject to drastic breaking changes moving forward.

🐛 Bug fixes

• Disable gRPC keepalive pings that trigger Quad disconnects (StartNotifyStates stream traffic keeps the connection alive, even in a standby state) @steamEngineer (#132)
• Recover gRPC exec after stale preflight tokens following standby idle @steamEngineer (#134)
• Refresh gRPC entity availability on session connect and disconnect @steamEngineer (#135)

Beta.3 adds WARNING-level auth snapshots on exec/preflight failures (including seconds_since_last_notify and cloud key expiry), so you can tell stale rolling tokens from expired Sony session keys without enabling debug mode. Turn on Verbose gRPC debug logging in integration options for the full trace: notify deltas, preflight/GetSessionRandom retries, exec tx/rx sizes, and session-restore retries.

📚 Documentation

• Add gRPC auth lifecycle reference and long-standby troubleshooting @steamEngineer (#134)

Full diff: v2.0.0-beta.2...v2.0.0-beta.3