Re: Yet another feature request for the web-server

[Shriram Krishnamurthi <sk@cs.brown.edu>]

MJ Ray wrote:

>						   However, for cookies
> to actually be impossible doesn't sound good.

Yeah, but they're not impossible for an *interesting* reason.  It
would be a simple matter to augment the document generation interface
to permit the specification of such headers.  

Francisco, I think allowing optional header information would be a
much better interface than exposing the port directly.  (Eg, the mail
libraries let you do this kind of thing, separating headers from the
message body.)  Once you expose the port, you can never assume
anything about its use -- eg, while you're trying to write one thing,
the user may concurrently write something else.  It seems better to
gather up all the information and localize the stateful operation.

Shriram