Breaking: Change how interface node gets converted (fixes #201)

[soda0289]

This commit create an TSInterfaceDeclration and TSInterfaceBody node when converting typescript interface

[eslintbot]

LGTM

[azz]

Do we need superTypeParameters for interfaces too?

interface B<C, D> { }
interface A extends B<string, number> { }

Never mind, it's covered by TSInterfaceImplements's typeParameters.

[soda0289]

I should change my tests to use extends. I thought interfaces implement other interfaces, but that's a mistake.

Maybe change the nose type to TSInterafceExtends?

[JamesHenry]

It's not a parse error, but this is invalid TypeScript

Interface declaration cannot have 'implements' clause.

[JamesHenry]

See comment about this being invalid TypeScript

[JamesHenry]

When this is updated to use extends, we could go for superInterface to align with superClass? There is no precedent in flow AFAIK...

[JamesHenry]

On the whole, I definitely like the change, we may even want to look at refactoring the body in a future PR. Flow uses ObjectTypeAnnotation for that and a few other things. Not sure that existed when we started this

[azz]

When this is updated to use extends, we could go for superInterface to align with superClass? There is no precedent in flow AFAIK...

Can't. Unlike a class, an interface can extend many other interfaces or classes.

https://astexplorer.net/#/gist/5ef8ee37c06a2390c37aae0f9a61ed04/bf68403565ba586b351e1a46c73d1a4232c6cdca

[JamesHenry]

How about heritage: TSNode[]?

Then we could even capture the semantically invalid implements examples in the AST, as TypeScript itself would...

[soda0289]

I'd be fine with using heritage. I'll also change the heritage node to TSInterfaceExtends but it could be TSInterfaceHeritage.

[eslintbot]

LGTM

[soda0289]

@JamesHenry I changed implements property to heritage.Also changed the node type to TSInterfaceHeritage

[eslintbot]

LGTM

[JamesHenry]

I'm pretty happy with this to go in now, thanks @soda0289! 😄

Any other thoughts @azz?

[azz]

LGTM 👍