Re: HTTP/QUIC Diverging from HTTP/2

> On 10 Mar 2017, at 02:09, Mike Bishop <Michael.Bishop@microsoft.com> wrote:
> 
> Summary of feedback on this so far:
>  • Ekr:  “Hoping we can converge, or at least maximize overlap”
>  • Stefan:  “any hope to [converge] needs to be abandoned”
>  • Ian:  “not excited… but it may… be the right thing to do”
>  • Patrick:  “separate protocols”
>  • Martin:  “keep the differences minimal” but “we're really building a new protocol”
>  • Alcides:  if it’s different, make the differences clear
>  
> If I’ve misconstrued anyone’s response, please speak now; and obviously, more opinions are welcome.  I would also appreciate reviews on the text of the PRs themselves.
>  
> Unless there’s strong pushback in the next ~14 hours, I expect to incorporate both PRs tomorrow, prior to the -02 publication.  As Mark noted recently, that doesn’t claim full consensus has been reached, but there appears to be general support for considering these separate protocols which are closely related, rather than two variants of a single protocol.

FWIW, I’m somewhere in between Ekr and Stefan. I think that where possible overlap and convergence should be pursued, but I don’t think we should hold our breaths. Ultimately I think in software implementations there will be very little common code between HTTP/2 and QUIC: barely any more than there is between HTTP/1.1 and HTTP/2, and potentially less. That suggests that HTTP/2 and QUIC are no more closely related to each other than HTTP/2 is to HTTP/1.1.

That sounds to me like “close cousins, but different protocols”.

Cory

Received on Friday, 10 March 2017 08:39:37 UTC