Michał Muskała

The rule I usually follow: 1. Define protocol in it's own file 2. For stdlib types or types from dependencies define the implementations in the same file 3. For custom...

Yeah, I don't think I have anything to add over what I already wrote here.

We were discussing this issue briefly on IRC. There's no 100% sure way to detect if we're in umbrella, but there are couple heuristics that could be used: - checking...

Would it be reasonable to have this as a default in rebar? If it's going to be opt-in, nobody's going to opt into it, because there's no direct benefit for...

I think the behaviour I'd prefer is that on the newline when I start typing the name of the function, it offers to complete, so rather something like: ``` foo(A,...

My reasoning is that this is in general how completion works - you start typing the name, and it's completed for you. Doing this for clauses seems very natural to...

I don't think edoc supports documenting types in the first place

Not directly, but it could make it simpler. The parser tries to keep as much information as possible (e.g. retaining full AST for values of attributes instead of converting to...

I don't have plans to work on this soon-ish, I might if it becomes frustrating enough, though 😉 If I pick it up I'll ping this issue.

FWIW I think much bigger issue with `rebar3_format` is https://github.com/AdRoll/rebar3_format/issues/216, which can lead to the code changing semantics in a very subtle, and hard to spot way