ietf-openproxy
[Top] [All Lists]

HTTP/OCP: short-circuit

2003-10-23 10:14:38

On Thu, 23 Oct 2003, Martin Stecher wrote:

First here are some thoughts about short-circuits in the request-profile. I 
think handling this one is not so complicated and can be added to the 
existing draft:

  Before the callout server sends back any application message part,
  it MUST decide whether it wants to adapt the request or to
  short-curcuit. It will either send back request-header,
  request-body, request-trailer parts or response-header,
  response-body, response-trailer parts. Some parts could be
  optional due to conditions defined below but it MUST NOT mix parts
  of these two lists; on violation client MUST terminate
  transaction.

I agree with the above in principle. If we make any changes to the
what-parts-to-send-or-skip mechanism, the above may need to reflect
those changes, of course. As of now, the above together with adding
adapted response-* parts to request profile should suffice.

Alex.


<Prev in Thread] Current Thread [Next in Thread>