Re: Non-order processing in persistent connections

Josh Cohen:
>
>Thanks for your comments.   Out of order responses in HTTP is a topic
>which has come up more than a few times.  I beleive that most people
>agree that out of order responses or transacation identification would
>be a good thing.  However, specifying exactly how this would behave
>isn't a simple task.  I think that because of this, the consensus is
>that it is beyond the scope of the 1.1 version of HTTP.  
>That doesn't mean the future versions of http wouldn't have this,
>but http/1.1 is "feature complete" at this point.

Yes, going beyond pipelining was generally considered too complex to
put in http/1.1.  

I'd like to add though that, as far as I know, most people do not
consider out-of-order responses to be the logical next step in
improving 1.1.  The preferred next step seems to be to add
multiplexing of several HTTP connections over a single TCP connection.
Multiplexing has some additional advantages over out-of-order
responses.  See http://www.w3.org/Protocols/MUX/ for one effort in
this direction.

Koen.

Received on Tuesday, 19 May 1998 00:52:50 UTC