KVScheduler TODO

KVScheduler

🆕 New

• consider changing value status FOUND to DISCOVERED, found implies that we were searching for even though we didn't expect it and simply discovered it during Retrieve
• consider changing value status RETRIEVED to avoid disambiguation with Retrieve operation (dump previously clearly made distinction), ideas: NATIVE/AUTOMATIC/IGNORED/GENUINE/OBSERVED

Code changes

• safe/testing mode
• resolve unknown origin also based on dependencies (or also remove SB values when needed)
• derived properties as attributes, perhaps generated by the Model itself
• re-work graph to use the graph history rather than the run-time state (which is limited to the current time)
• display unsatisfied deps in the graph (needs to do the item above first)
• change to CRUD
• create distinction between immediate failure and total failure (after retries)
• consider removing IsRetriableFailure (or add references to retries)
• if Modify is not implemented acts as ModifyWithRecreate always returns true
• remove 4 fields related to prefix/keys

Documentation

• add example diagram of planned/executed operations
• describe dump (read)
• describe why is DumpDependencies needed
• add code examples for every field
• create example with custom descriptor
• control flow diagrams