Arrow function can be non intuitive for JS developers about civet HOT 6 CLOSED

@lorefnon TypeScript doesn't allow Type annotations on paren-less single parameters (another reason I choose to disallow them). There is definitely room to improve error messages in the LSP and I'd be happy for you to take a look.

The recommended Civet way to type annotate your example is:

z : (n: number) => number := &+1

Note the annotation is for the type of z to the left of the const assignment :=.

In cases where the ampersand function is used inline TS can usually infer the type correctly.

declare const x : number[]

y := x.map &+1

@edemaine in light of these examples I think that allowing inline type annotation of ampersand functions makes the intent a bit less clear but I'd be happy to look at some counter examples.

Also, I want to say thank you both for your contributions! It really helps me work out the practical details of the language to see how people feel most comfortable using it.

from civet.

I feel the "one arg optional parens" "feature" in ES is a design error. Civet also provides a better way to omit single arguments with the &+1 => $=>$+1 function shorthand.

You can use y := &+1 for your example above.

In cases where CoffeeScript did a feature earlier / better I prefer to use that design to "fix" things where ES left inconsistencies.

A lot of words to say this is by design but maybe a "civet esCompat" directive could include this for people who have a large legacy ES codebase to migrate.

from civet.

Ok, I'll close this since this design was intentionally chosen.

I can get used to always needing parenthesis for fn args.

The & syntax is not convenient other than places where the the type can be inferred because something like y := &+1 results in a ts error that the (mysterious 😄 ) parameter $ implicitly has an any type.

from civet.

@lorefnon Good point about the type inference. You could add a type annotation to the declaration side like: y : (n:number) => number := &+1 but I'm open to exploring a better way.

from civet.

@STRd6 I don't have a strong opinion/suggestion here, but I'll see if I can do something on the lsp side to make the error better. The current errors are not very obvious to someone who is not aware of the paramless arrow fn syntax.

from civet.

Would &: number + 1 or (&: number) + 1 be interesting typed shorthand for ($: number) => $ + 1?

from civet.