Re: client-store interface

On Sun, 2 Nov 2003, Robert Collins wrote:

[regarding 100 COntinue]
>
> We do? For all three cases? (1.0->1.1, 1.1->1.0, 1.1->1.1). 1.1->1.0
> gatewaying is the silghtly tricky one.

100 Continue is basically a hop-by-hop thing. There is no gatewaying
requirements or complications in implementing this in a proxy.

Once the client-side sees there is someone asking for the request entity
it should send 100 Continue if the HTTP/1.1 client indicates it is waiting
for one. You do not need any additional indications on the client side.

On the server side you need to implement 100 Continue logics, making the
server side indicate it wants a 100 Continue (with all the fallback
logics to not wait forever). Once the 100 Continue negotiation with the
upstream server have continued start forwarding of the request entity
(which will trigger the client side 100 Continue by the simple fact that
someone is requesting the request entity).

> > But I disagree that it should be the broker who tries to solve the "odd"
> > cases. These belongs in the server side processing as far as possible. The
> > server-side has far better knowledge of what can be done within the
> > protocol to solve the situation.
>
> Hmm. I'm -not- making the broker solve the odd cases. I'm giving the
> client the ability to do so, without breaking abstraction.

And I favor having the server-side solve the odd cases, for the plain
reason that it has much better knowledge of what the server-side is
capable of for the gatewayed protocol in question.

Regards
Henrik
Received on Mon Nov 03 2003 - 04:50:08 MST