Understanding the context of the cursor point about merlin

[let-def]

a) Merlin tries to guess the context under cursor in different way.
The normal path is if both parsing and typing succeeds normally (code is well-typed), then it will have full information about context (including local scope of expression).

If typing fails, it fallbacks to a "relaxed" typer that accepts type errors but tries to produce as much meaningful information as possible. While already available, some improvements not yet published have been made to further "relax" the typer.

If parsing fails, this is the hard case… Merlin is not able to do anything. Wrt to ocaml compiler, the parser is much more tolerant (it will report errors and tries some heuristic to continue parsing).
But this is not enough, I have some idea about possible improvements, I'll keep you informed about that.

b) Even if those information are available inside merlin process, they are not made available to the editor. I am not sure how to present these information to the end-user.

[roshanjames]

Thanks for the explanation.

About (b): I was thinking that one could ask auto-complete to say display the usual auto-complete with M-/ (or C-TAB as you have in the default bindings) and then then hitting some other key will show only the completion candidates that match the context type.

[let-def]

Implemented some time ago.