FW: Segmented recording to reduce data loss during crashes #147
Description
Proposal: Implement Chunked / Segmented Recording to Reduce Data Loss
Many action cameras use a segmented recording strategy: instead of writing one long continuous file, they record in fixed-length chunks (e.g., 1–5 minutes). When a segment reaches its time or size limit, the device immediately starts writing to a new file.
This reduces data loss in case of unexpected failures (e.g., crashes, power loss, storage errors), because only the last incomplete segment is at risk, instead of the entire recording.
Why This Could Be Useful for OpenEarable
From what I know, OpenEarable currently stores long continuous recordings, which can lead to substantial data loss if the device crashes before the file is properly closed. Using a chunked recording approach could provide:
- Improved robustness: Only the current segment is affected during a failure.
- Better reliability for long recordings: Avoid overly large files that stress the filesystem.
- Faster recovery: Finished segments remain valid even if the last one becomes corrupted.
Open Questions / Discussion Points
- What segment length is optimal for sensor data recordings?
- How should file naming be handled to maintain order and clarity?
- Is this even necessary or does our file type already ensures minimal data loss?