WebSockets

[mnot]

This issue is a placeholder / tracking issue for anything related to WebSockets support in HTTP/2.

[mnot]

Note that our charter says we need to "will coordinate this item with...[t]he HYBI Working Group, regarding the possible future extension of HTTP/2.0 to carry WebSockets semantics."

[mnot]

See also:
http://tools.ietf.org/html/draft-hirano-httpbis-websocket-over-http2-00

[mnot]

And:
https://github.com/yutakahirano/ws-over-http2/blob/master/ws-over-http2-message-mapping.md

[mnot]

Discussed in NYC; we believe this could be done with extensibility (e.g,. with a new ALPN token, maybe new frame types + settings) - but we're not going to explicitly design it.

[martinthomson]

@thedrow we haven't decided not to do websockets, we've only decided not to block HTTP/2 until a design for websockets over HTTP/2 is done.

[lenovouser]

Are there any updates on this? It has been a whole year now since someone commented on this, are WebSockets still going to be implemented in HTTP/2?

Also see caddyserver/caddy#189

[MartinKei]

What's the status on this one?

[martinthomson]

Awaiting someone to do some work: implement it, convince others to implement it, etc...

[AlanQuatermain]

I'm looking into this for the HTTP/2 upgrade across our systems at LinkedIn. I'm given to understand that the lack of WebSockets makes this upgrade more arduous for us (i.e. minimal benefit unless we roll out these changes across a very wide number of micro services across a large number of colocation locations, etc.) I've found Yutaka Hirano's draft spec at the IETF and it looks like I can take that & run with it fairly simply.

My question here is: who else is interested in this? Has there been any other work in this direction since Hirano's last draft nearly three years ago?

[bagder]

@AlanQuatermain: there's now a new draft being discussed and worked on in the HTTPbis WG: https://datatracker.ietf.org/doc/draft-mcmanus-httpbis-h2-websockets/