The SDK retries a request that fails before its response starts, but not a
block body cut off part way through. That error stopped playback or failed
a download, and the SDK's new keep-alive pings make it the way a dead
connection now ends: seconds in, rather than at the storage timeout.
A block is now fetched up to three times, 250 ms and then 500 ms apart,
while ProtonError::is_retriable says a replay could succeed. A permanent
refusal still fails on the first attempt.
- the streaming cache's budget (1-16 GiB) is chosen in Settings, shows
what it holds, and can be emptied; a lower budget evicts at once
- hardware decoding can be turned off for a driver that shows green
frames, from the next file
- both live in desktop.json rather than the playback prefs shared with
Android, whose bridge record would reset them on every save. A file
that will not parse is never overwritten
A block the player waits for no longer waits on its disk-cache write and
two fsyncs; the put runs in a spawned task and only the block itself is
synced before the rename. Opening a stream can fetch the first and last
block together, where MKV cues and the MP4 moov live. Downloads open a
copying stream that reads the disk cache but never fills it.
crawl_reporting reports the running file count after each listing.
nvrl