Scott/Walden,
It was my understanding that the "doug server" in this situation was
actually going to be a minimal implementation of the HTTP protocol. Was
I mistaken on that part?
Not even very much of the HTTP protocol -- just handle the two specific
requests (presumably sent by xmlhttprequest) and return a response.
If it only needs to handle one particular request that will always be
formatted the same way, this would only be slightly harder than a
custom/proprietary socket program.
Two particular requests, formatted the same way. If I go that route, it
would just be listening for the request, a select case to decide how to
handle it (with only two current branches), and spawning or whatever the
task to send the response. The OTHER clause would at most log unrecognized
requests for either debug or audit purposes, but definately would not be any
real HTTP server implementation. (My response would of course have to
generate the appropriate HTTP headers to prefix my XML data, but that is
hardly a full HTTP server implementation.)
It is the hosted website which would run the real HTTP server (and I don't
know if it is classic ASP or ASP.NET), and it would presumably use
xmlhttprequest to send requests to me. I don't think it could / should know
the difference between a "doug server" providing the response and Apache
using CGI to call a RPG ILE program to generate the response.
That said, if the client will agree to run Apache, I will suggest that
route.
Doug
midrange.com
