We have different classes of errors/irregular situations occurring in QM and QM clients, including at least:

• errors in disk data (e.g. corrupted, missing or unexpected files) (which may or may not be caused by errors in the implementation)
• IPC protocol errors (caused by a malicious or malfunctioning peer)
• syntax errors in user input or bad API usage
• unexpected failures of supposedly near-infallible operations
• internal inconsistencies
• aborts, e.g. during shutdown

It would be good to be able to report errors differently for the different classes. Also, a consistent way of handling/propagating/crashing/suppressing errors in these cases should be defined and implemented.

Yeah, so we will be able to filter telemetry data based on that or even suppress reporting to telemetry for some error classes.