Add TypeNode#has_method?

[Sija]

Fixes #4472

[bew]

Nice addition, you should add some specs for this

[Sija]

@bew there aren't any for the rest of related methods, I can add it though.

[bcardiff]

@Sija yes, please let's add some spec in /spec/compiler/codegen/macro_spec.cr.

I would suggest also to split the fixes in the docs in other PR 🙏

Regarding the functionality, I have some doubts how it should interact with method_missing. Currently responds_to returns false if no method with that name is invoked somewhere, which is reasonable given that the types are fixed after the compilation finishes. But during macros evaluation, depending the evaluation order the has_method might return true if later a method_missing generates that method.

[asterite]

You can simply use the name variable here, there's no need to traverse all metadatas

[asterite]

@Sija Ping :-)

[Sija]

BoolLiteral.new !!type.defs[name]? will do?

[asterite]

Yes, or type.defs.has_key?(name)

[asterite]

Actually... type.has_def?(name)

[Sija]

done :)

[bew]

I think that BoolLiteral.new(type.has_def?(name)) is enough, as Type#has_def? is already returning a boolean

[Sija]

@bew nope, returned type is Bool | Nil. I'd say it should indeed return just Bool, yet it might be out of the scope of this PR.

[bew]

right, sorry about that (found why: some_hash.try &.has_key?(some_key) can be Nil with the try..)

[Sija]

@bcardiff I've added specs as per request. I'm not sure what to do regarding method_missing and macros evaluation order though... any ideas? /cc @asterite

[asterite]

@Sija don't worry about that, it's a separate issue

[Sija]

Is it GTG? Travis CI xcode8.2 build failed, rest 's good.