Re: Our Schedule

On Tue, May 27, 2014 at 12:04:54AM +0200, Martin Nilsson wrote:
> On Mon, 26 May 2014 19:53:21 +0200, Mark Nottingham <mnot@mnot.net> wrote:
> >
> >The previous discussion that Patrick was referring to has a lot of  
> >background.
> >
> 
> I can go further back in time and say that we considered headers to be a  
> problem both to get requests ouy quickly in Opera Turbo and that headers  
> sometimes created a significant overhead in the response path. We use a  
> much simper compression thouhg. Both Opera Mini and Opera Turbo uses delta  
> compression on header level and Mini also uses static header name table  
> (kind of) and typed values (i.e. you can send a date or a size as an n  
> byte integer).

Yeah so we're back to the presentations Roy made about Waka or that I
made about our closely related proposal of typed encoding. Someone also
sent a link about the way WAP used to encode headers to reduce them. I
think that many people came to similar designs without even being in
relation because the first thing that strikes is the large header field
names for few used values, and the repetition of similar header values
between requests.

>From a technical point of view, I'd also like the header compression to
be optional, but I think that's a no-go as it will waste an RTT in the
negotiation. However we might possibly consider having a simpler variant
of HTTP/2 that does not implement header compression and which is
advertised in the TLS or upgrade handshake (eg: http/2-). That could
be useful for simple devices which don't need all the bells and whistles,
and would not cost an extra RTT.

Willy

Received on Tuesday, 27 May 2014 05:46:06 UTC