Re: HTTP 2.0/flag day

On Tue, Mar 27, 2012 at 11:47:39AM +1300, Amos Jeffries wrote:
> On 26.03.2012 22:13, Ted Hardie wrote:
> >I'm seeing a bunch of messages that seem to be based on the 
> >assumption
> >that after the creation of an HTTP 2.0 that all HTTP 1.x would
> >disappear in the transition to HTTP 2.0.  I am not anticipating a 
> >flag
> >day here, and I would guess that HTTP 1.1 would still be around for
> >any use cases that do not need features of any 2.0 proposal.  Does
> >that make sense to others, or do we have a design constraint that all
> >use cases currently met by HTTP 1.X must also be met (with the same
> >security and performance properties) by 2.0?
> 
> By my reading that is the chartered requirement anyway. If we do not 
> meet or improve all scenarios already *in-use* within HTTP/1.x then the 
> upgrade attempt will have failed. Why bother moving to or even creating 
> a protocol that does not do what we need it to?

I agree with you Amos. If the new protocol has no chance to progressively
replace 1.1 just like 1.1 slowly took over 1.0 and 0.9, then it will be a
mess because instead of having to support 3 versions in our products forever
we'll have to support 4. There will be a transition period for sure, but
with enough incentive to migrate, maybe we can get rid of most of 1.1 and
below in less than 10 years (at least I hope).

> IMHO, best not to do it with a flag-day, but the WG may disagree on 
> that.

I don't believe in flag days for this. If we have something compatible
with what exists, it will be slowly but surely be adopted.

Willy

Received on Tuesday, 27 March 2012 04:55:45 UTC