should we mimic the R6 API more?

[paciorek]

This came up in my look into #101.

I'm not proposing it, just throwing the idea out.

Do we want to use the self$ and private$ style referencing to member vars/methods as R6 does in nClass methods? I.e., code would look like localVar + self$memberVar rather than localVar + memberVar?

[perrydv]

Yes. Quick thoughts:

• In R6 the need for self depends on portable and I'm not sure what other issues would need to be addressed in relation to that.
• At some point we can add Rprivate and Cprivate, hoping those will be relatively simple to add.
• Supporting self is relevant for ability to drop user into a debugging session upon unexpected behavior #99, where it would be nice to include "self" in the list of variables to access to from the browser in R.