W3C home > Mailing lists > Public > ietf-http-wg@w3.org > October to December 2013

Re: Non-browser uses (was Re: disabling header compression)

From: Howard Johnson <hj@BridgeportContractor.com>
Date: Fri, 13 Dec 2013 14:57:16 -0800
Message-ID: <52AB90CC.2010100@BridgeportContractor.com>
To: HTTP Working Group <ietf-http-wg@w3.org>
At it's very core is not the current issue at hand:  just how to 
gracefully move from http1.1 to http2.0?  In other words, without a lot 
of time consuming negotiation on every request, as systems all over 
earth gradually adopt 2.0 and beyond.  And yes, we are trying to avoid 
something similar to the 1.1 gzip negotiation slowdown problem in the 
process.


I liken this to the transition from when we only had telephones (think 
http1.1) to when we email began (think http2.0).  How did we make that 
transition in communication protocols?

First we never got rid of telephones, they just got used less and 
differently.  (http1.1 will be here for awhile.)

As email started to appear (in our case http2.0) I remember that a lot 
of time was spent finding out if a particular friend (think: a server) 
had email or not.  Today the question is, does a site support http2.0 or 
not and with what features?

Now one could negotiate every single time we needed to communicate a 
message, in other words, first call by telephone and ask if the person 
had email, and if so send via email, but that's not what we did for it 
was too inefficient of a method of protocol transition.

Instead once we discovered a friend with email we noted that fact (kept 
a personal address list).  We also created public directories of who now 
had email.

I think this is how we should approach the transition from http1.1 to 
2.0, and from uncompressed headers to compressed headers, and other 
possible features, etc.  Provide the infrastructure to discover 2.0 
(both by negotiation and by public directory).  Let clients and servers 
use this cashe of information to decide how to most efficiently send 
http.  In other words negotiate once if need be, but then remember what 
the result was, and henceforth blindly send 2.0 and/or compressed 
headers to those we know can accept it.

Via this approach I think it will be in every bodies best interest to as 
quickly as possible adopt 2.0, just as we adopted email back in the day.



Received on Friday, 13 December 2013 22:57:41 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 11:11:20 UTC