Re: HTTP 1.1 As Universal Transport?

>But does this makes it possible for a server to send events to
>a client. I think not.
>Very sad.

A server initiating a communication with a client is rightly outside the model. 
The only distinction between clients and servers in client-server architectures 
is that a client initiates while a server passively accepts. Thus if a "server" 
starts initiating actions it isn't acting as a server, its acting as something 
else and to the recipient of the communication it looks like a client.

The idea pushed about two years ago was to build in a server into the browsers 
which could then be used to accept notification. I still think this is the best 
route to go. It means writing no new specifications.

The route of writing some mux layer which allows a server to send asynchronous 
messages to the client once a communication has been established appears to be 
prommising but it is probably best persued in the context of http-ng and not in 
the context of rfc-822 style http. This is not the time for that discussion. 


Received on Thursday, 16 May 1996 09:05:15 UTC