Support for CLOSE_FRAME

[aboba]

We now have: https://datatracker.ietf.org/doc/draft-ietf-quic-reliable-stream-reset/

What changes (if any) should be made to the API to support this?

[bashi]

Probably the IETF draft needs to be revised to specify the behavior about CLOSE_FRAME first, then the W3C spec refers the behavior? cc/ @vasilvv

[vasilvv]

Well, there are two questions:
(1) Use of CLOSE_FRAME internally by WebTransport itself. That would have to be added to the IETF draft.
(2) Providing an API for Web applications to use it. This should not be that hard (though the async nature of stream writes does make it somewhat tricky to do exactly), but I'd wait until someone actually asks for that.

[bashi]

+1. IIUC CLOSE_FRAME is useful to associate streams to a WebTransport session reliably. I'm not sure there are concrete use cases for applications.

[aboba]

For something like partial reliability (e.g. setting a timer and sending a reset if it expires), I don't think reliable reset makes sense. If you're giving up on a frame, you don't care if part of it is sent reliably. So it wasn't clear to me why MoQ would need reliable reset.

[jan-ivar]

Meeting:

• Wait and see
• Wait for underlying concept to materialize