>We have a 'host' that runs at corporate, which processes transactions from the field (also using FTP). The idea I had, crazy or not, is that when a remote user who wants to run a report against a large historical data file would have his/her 'front-end' app send nothing more than a query expression string to this same 'host', which would run the SQL statement based on the query, and FTP the resulting recordset back to the remote (which would be polling a folder on the FTP server, waiting for the 'host' to actually send it).
>

Hey Kevin-

what you are trying to do is done pretty commonly, and what is really needed is a client side app that uses HTTP to get results from the server. Using HTTP instead of FTP, you're not polling, but keeping with true client/server, request/response architecture.

THe first thing I would do if I were you is to download Rick Strahl's wwIPStuff. It has the methods you can use to pass data to and receive results from a web app. The perfect man for the transport job, IMHO, is XML. Also download wwXML from Rick's site. It has two methods that will be very useful to you: CursorToXML and XMLToCursor. Using these tools, you can run a query on the server, convert the resulting cursor to an XML document, return that document as the results of your request, have the client parse the XML and create a cursor from them and Voila, you're done. Now you have VFP data on the Server, and VFP data on the client, and you used XML in between. Once you have your framework in place, everything is a breeze. I have a couple of apps using this strategy in productoin now, so feel free to bug me about the details.

BTW, your proposed solution isn't that bizarre- I have built similar working apps, but never will again now that I am comfortable doing it this way.