Sort out dataVersion in ConcreteDataAttributePath

[yunhanw-google]

I assume this was the result of discussion with @mrjerryjohns but I am not sure this makes sense. ConcreteDataAttributePath is used for both writes and read responses. And while a read response always has a data version (except not, if there is an error!), a write might not. And we don't have provisions for indicating "no data version" here.

Are we just ignoring this member (but still paying its cost) in the write code? That's a little confusing, and why are we passing the data version in the path for reads and out-of-band for writes (but still have this thing in the path for writes)?

Not going to block the PR on this, but please make sure a followup is filed to fix this. Maybe we need to have separate path representations for read responses and writes (e.g. the one for read responses could inherit from ConcreteDataAttributePath) but also include a dataversion.

Or maybe this should in fact be an Optional, since for reads this value may not be available either? And then we should just use it from the path in the write case?

Originally posted by @bzbarsky-apple in #14711 (comment)

[yunhanw-google]

#16602 (comment), prefer having have a ConcreteDataWriteAttributePath (optional dataversion) and a ConcreteDataReadAttributePath (non-optional data version).

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.