add WebUSB WebIDL

[MarcAntoine-Arnaud]

issue #1694

[alexcrichton]

Thanks for this! We in general only add stage 3 and beyond proposals to this repository. I can't seem to find documentation, but do you know at what stage in the standards process this proposal is at?

[MarcAntoine-Arnaud]

I don't know where we can find the stage of the proposal...
But this patch don't work, I cannot access to USB feature (in the documentation generated with: cargo doc --all-features. Do you have any idea why ?

[Pauan]

@alexcrichton I don't know what stage it is, but it is marked as Experimental on MDN, and only Chrome and Opera have support for it right now.

In addition, MDN says that it's currently at Draft status, so it's definitely not stable.

And the status section says this:

This specification was published by the Web Platform Incubator Community Group. It is not a W3C Standard nor is it on the W3C Standards Track.

[Pauan]

I wonder if maybe we should have support for unstable APIs, perhaps behind an unstable-api feature or something like that.

And we would reserve the right to break anything behind unstable-api at any time, for any reason, even in minor releases.

[alexcrichton]

Ok so it sounds like we shouldn't add this as-is because it's not an official spec yet, but I like the idea of having experimental specs here that we can break at any time. Prior art for that I'm aware of isn't to use Cargo features but rather RUSTFLAGS=--cfg=web_sys_unstable_apis or something like that.

That way it's a bit more experimental in that it forces consumers to always compile with the right flags. How's that sound to enable this?

[Pauan]

How's that sound to enable this?

That sounds reasonable to me. It means that libraries will have a harder time relying upon unstable features, so they will be mostly used by applications, which is how it should be.

[MarcAntoine-Arnaud]

@Pauan @alexcrichton thanks for comments.
I'm agreed it make sense to put it in unstable api.

[alexcrichton]

This hasn't seen movement in quite some time now so I'm going to close this, feel free to resubmit it though!

[alexcrichton]

FWIW I've opened #1950 to track longer-term about how we plan to support unstable WebIDL files.