What is the purpose of SETTINGS_ENABLE_WEBTRANSPORT in h2?

[DavidSchinazi]

I can't seem to remember or work out what the purpose of SETTINGS_ENABLE_WEBTRANSPORT is now that we have SETTINGS_WEBTRANSPORT_MAX_SESSIONS. This leads to the following questions:

• If the server sends SETTINGS_WEBTRANSPORT_MAX_SESSIONS>0 but SETTINGS_ENABLE_WEBTRANSPORT=0, what does that mean?
• If the client sends SETTINGS_ENABLE_WEBTRANSPORT=0 but sends a WebTransport request, what is the purpose of having the server fail the request?

[martinthomson]

Yeah, I thought that we would remove SETTINGS_ENABLE_WEBTRANSPORT when the session limit was added. If nothing else, it removes those awkward interactions.

[DavidSchinazi]

Chair: discussed in editor's meeting, sounds like we should remove SETTINGS_ENABLE_WEBTRANSPORT. We might add another setting to be removed by the RFC editor for versioning between drafts

[ekinnear]

When an HTTP/2 connection is established, both the client and server have to send a SETTINGS_ENABLE_WEBTRANSPORT setting to indicate that they both support WebTransport over HTTP/2.¶

The server also needs to send a SETTINGS_WEBTRANSPORT_MAX_SESSIONS setting with a value greater than "0" to indicate the number of concurrent sessions it is willing to receive.¶

IIRC, we discussed at IETF 115 and only the server needed to send MAX_SESSIONS, so we kept both. Can spell that either way though.

[ekinnear]

If we're doing this for H2, should we also do it for H3, since we have SETTINGS_MAX_WEBTRANSPORT_SESSIONS there, too?

[DavidSchinazi]

• h3 requires SETTINGS_ENABLE_WEBTRANSPORT because it changes the format of server-initiated bidi streams
• h2 has no such requirement for a setting