Ingvar Stepanyan

> Standard descriptors could appear in parsed form while non-standard descriptors could appear as a generic "descriptor" object with the type and length parsed out but the descriptor body as...

> My only concern is that we need to double-check that we can get the USB speed on all platforms. FWIW libusb has "unknown" constant for this scenario. It should...

> > The bMaxPacketSize field reports the maximum packet size for endpoint zero. All devices must support endpoint zero. I suspect that definition is somewhat outdated. It was true for...

Can't find a good reference link, but this [patch discussion](https://linux-usb.vger.kernel.narkive.com/wSEOXlqR/patch-usb-composite-fix-bmaxpacketsize-for-superspeed) that pops up among first Google results is helpful: > For bMaxPacketSize0 we usually take what is specified in ep0->maxpacket....

I don't have specific usecase where this would be necessary, but yes, BYOB could be interesting for performance reasons as well as for simplifying Wasm bridge code. One other thing...

> If it is changed to add the `.rs` extension after the `filename_hash` function call, it will be resolved. But I am not sure the file produced is valid rust....

I think in the face of Tyler's proposal that changes identity/address representation altogether, this PR might become obsolete.

Same for `BytesSource` and maybe more types?

This needs rebase now that Identity/Address representations have changed + snapshot regen.

@reillyeon https://github.com/WICG/webusb/issues/25#issuecomment-208661215 seems to suggest that all major platforms support cancelation? Can we always do whatever `libusb_cancel_transfer` does?