Re: Multi-GET, extreme compression?

[Mark Nottingham <[e-mail]>]

On 19/02/2013, at 3:26 PM, Nico Williams <[e-mail]> wrote:

> On Sat, Feb 16, 2013 at 12:01 PM, Phillip Hallam-Baker <[e-mail]> wrote:
>> HTTP 1.1 has a request/response pattern. This covers 90% of needs but means
>> that if the protocol is followed correctly forces a round trip delay on each
>> content request. Which of course leads to various browsers pushing the
>> envelope and pushing multiple requests out before responses have come back.
>> 
>> With content streams this is not necessary of course... In fact that is
>> pretty much the purpose of having streams.
>> 
>> Which suggests a need for a Multi-GET method to allow a request for a list
>> of content...
>> 
>> If we had such a method then the format would be something like
>> 
>> MGET <Common Headers> List <URI, Content header>
>> 
>> And the typical communication pattern of a browser would be:
>> 
>> GET /toplevel.html
>> MGET </image1.jpg /image2.jpg ...>
>> 
>> Given this particular communication pattern which has an implicit delta
>> encoding, do we really need to worry about a separate delta encoding?
> 
> The problem here is that the user-agent needs to get the top-level
> resource first, then it will know the names of the other resources.
> We can probably do better.

Nico,

If I understand you, you're talking about making some really fundamental changes to the Web Architecture, which is squarely out of the WG's charter.

I don't mind discussing ideas and understanding how we got here, so long as they don't distract from our work. I get the feeling that this is starting to happen. 

Again, if you have a proposal, please write it up in detail and make it to the WG; endlessly discussing the minutia of a half-formed idea is not a productive use of anyone's time.

Thanks,


--
Mark Nottingham   http://www.mnot.net/