Re: [hybi] moving forward on HTTP extensions

[Salvatore Loreto <salvatore.loreto at ericsson.com>]

Pieter Hintjens wrote:
> On Thu, Jun 18, 2009 at 1:24 PM, Salvatore
> Loreto<salvatore.loreto at ericsson.com> wrote:
>
>   
> > we have tried to list all the issue on the Best Practice draft:
> > http://www.ietf.org/internet-drafts/draft-loreto-http-bidirectional-00.txt
> >     
>
> This is an excellent text and immediately useful to me, and I'll
> reference it in the RestMS design.
>
> The only improvements I could offer are some language simplifications,
> the use of passive tense in parts of the text makes the explanations
> less precise than they could be.  I'd be happy to edit.
>   
any opinions on potential future work items for bidirectional HTTP that 
are listed in the draft?

it would be even more interesting hear your (as well from all the other 
people interested in this issue) ideas/opinions
on how HTTP could be extended to better support bidirectional communication

> Two questions that occur to me are:
>
> * how these best practices relate to RESTful use of HTTP
>   
good point, we haven't said anything about the RESTful use of HTTP for 
bidirectional / long polling.
It would be great explicitly adding some statement explicitly saying that:
- every URI designates exactly one resource
- only GET has to be used to retrieve a representation of a resource.

but I have to admit that I have not looked into RestMS yet!

> * how protocols like WebSockets and BWTP fit into the picture
>   
the best practices draft only relates to the technologies that provide 
server-push using HTTP.
Both WebSocket and BWTP are instead working on a different aspect of the 
server-push technique,
that briefly is upgrade from HTTP to a different bidirectional protocol.
So both the protocols do not fit at all into this draft.

cheers
Sal