Expose BigInt bindings

[RReverser]

PR #188 a while back added mapping of {u,i}64 to BigInts for custom bindings, but there is still no implementations of From<{u,i}64> for JsValue or raw BigInt bindings which makes their construction a bit cumbersome in generic contexts.

[RReverser]

Looking into this further, I see that some other types that are castable when creating manual bindings, are not supported with explicit conversion to JsValue.

I wonder if it would make sense to provide a blanket implementation of From for JsValue for any type supported by wasm-bindgen itself (via : JsCast constraint or something like it)?

[alexcrichton]

We've previously decided that any phase 4 proposals can be added to js-sys, so once it's there we're good to go!

[RReverser]

Ah I see. And how about a blanket From implementation for any types supported by wasm-bindgen itself as suggested above?

[alexcrichton]

I think we currently have that? Although we may be missing a few by accident!

[RReverser]

That's what I'm trying to say - BigInt seems to be one of these, if I understand correctly. I mean, it's possible to create custom function with #[wasm_bindgen] attribute that accepts/returns u64 and that would translate to BigInt on JS side, but it's not possible to similarly convert u64 to JsValue directly.

[alexcrichton]

Oh sure yeah makes sense, and seems fine by me to add!

[alexcrichton]

We decided awhile back that we'd bind stage 4+ proposals, and it looks like this is currently stage 3 so I'm going to go ahead and close this.

[RReverser]

Nooo, I'm working on it in a branch, there is just a bunch of different changes needed. ;(

[alexcrichton]

Oh that's still fine! I just figured we didn't need a tracking issue for something that's not quite ready yet, but there can always be a crate on crates.io that we pull in later

[RReverser]

Wouldn't it be a bit inconsistent that wasm-bindgen already internally relies on BigInt for u64/i64 support, but doesn't expose explicit bindings to it?

This was the main motivation for this issue.

[alexcrichton]

It only relies on it in an experimental fashion because it's the only way to work with some of those types.

[recmo]

BigInt has since moved to stage 4: https://github.com/tc39/proposals/blob/master/finished-proposals.md