Re: Identifying our deliverables

[Jana Iyengar <[e-mail]>]

I'll add to the chorus: I'm in favor of this change.

On Tue, Oct 30, 2018 at 9:36 AM Roberto Peon <[e-mail]> wrote:

> Since there isn’t any particular need to identify QUIC the transport
> (without mechanisms to explicitly multiplex multiple protocols), I think
> calling HTTP-on-QUIC H3 is the right move.
>
> QUIC without anything else is as meaningless as TCP without anything else
> from a what-the-server-will-speak perspective.
> -=R
>
>
>
> *From: *QUIC <[e-mail]> on behalf of Mikkel Fahnøe Jørgensen
> <[e-mail]>
> *Date: *Tuesday, October 30, 2018 at 7:29 AM
> *To: *Ian Swett <ianswett=[e-mail]>, Lucas Pardue <
> [e-mail]>
> *Cc: *Mark Nottingham <[e-mail]>, IETF QUIC WG <[e-mail]>,
> Willy Tarreau <[e-mail]>
> *Subject: *Re: Identifying our deliverables
>
>
>
> Recent edits in transport also use terms such as “This frame is only used
> in the QUIC layer” which is slightly confusing when QUIC/HTTP is also a
> (higher) level QUIC layer while “QUIC layer” refers to the transport layer
> only, implicitly.
>
>
>
> Mikkel
>
>
>
>
>
> On 30 October 2018 at 15.26.19, Lucas Pardue ([e-mail])
> wrote:
>
> I echo Ian's thoughts, "HTTP over QUIC" doesn't work well as a term in
> practice.
>
>
>
> I support this change, using an incremental number is logical, and is
> compatible with the ecosystem that expects an HTTP version number.
>
>
>
>
>
>
>
> On Tue, 30 Oct 2018, 13:56 Ian Swett, <ianswett=
> [e-mail]..org <[e-mail]>> wrote:
>
> From a completely practical perspective, I support this because I can't
> tell you how many times I've had to explain the difference between the HTTP
> over QUIC mapping and QUIC the transport.
>
>
>
> And moving the HTTP over QUIC and QPACK documents to the HTTP working
> group seems like the right thing to do.
>
>
>
> On Tue, Oct 30, 2018 at 1:21 AM Willy Tarreau <[e-mail]> wrote:
>
> On Sun, Oct 28, 2018 at 11:30:55AM +1100, Mark Nottingham wrote:
> > To address this, I'd like to suggest that -- after coordination with the
> HTTP
> > WG -- we rename our the HTTP document to "HTTP/3", and using the final
> ALPN
> > token "h3". Doing so clearly identifies it as another binding of HTTP
> > semantics to the wire protocol -- just as HTTP/2 did -- so people
> understand
> > its separation from QUIC.
>
> FWIW I always thought that HTTP/3 should be its final naming -- just like
> SPDY
> became HTTP/2 -- for various reasons, one of them being encouraging
> adoption
> by the protocol being presented as the natural upgrade to the previous ones
> and not a competitor that's been there for some time and suddenly comes as
> an RFC.
>
> H1 with 723x, then H2 with 754x have shown that it's not a problem to cover
> multiple areas with a common name and that the protocol itself is in fact a
> family of protocols and mechanisms. So QPACK will naturally fall under H3
> just like HPACK naturally belongs to H2.
>
> Just my two cents,
> Willy
>
>