We have a win32 application which uses a .Net DLL to make a single web 
request to an ASP.NET website hosted on IIS6. This application works fine at 
our end, but at one particular client's site, win32 application crashes when 
making the web request without giving any information about what could be 
causing the crash. Furthermore, I can't reproduce this at our end even using 
a similar stack as theirs. I was hoping to get some help on methodology to 
try to determine what could be causing the application to crash in this way.

This is what I have so far:

* The web request is made by the .Net DLL to an ASP.NET request handler, 
which both receives and returns an XML-serialised DataSet

* The application works as expected at our end and on other clients' sites, 
but not at this one particular client

* The application crashes both when making the request through their 
firewall, and to a local web server on the same internal network

* The application doesn't crash when the request is proxied through Fiddler

* The only difference between the content of the request when it's a direct 
request or proxied through Fiddler is that Fiddler doesn't forward the HTTP 
1.1 100/Continue response IIS sends to the initial request back to the client 
DLL - this is 'absorbed' by Fiddler

The win32 application contains exception handling code to handle any 
exceptions thrown by the .Net DLL, so I'm curious as to why it's bombing out 
like that. My approach to testing this would be:

* try writing a .Net wrapper for the .Net DLL to see if it might throw an 
exception giving more information as to what is causing the problem

* try using HTTP 1.0 to see whether the 100 response is somehow causing the 
problem (although I'm not sure why it should, both request and response seem 
well-formed).

Any other suggestions gratefully received.

Thanks,
Nick