Thank you for your reply.

I am familiar with the West Wind product you're speaking of. I have used Web Connection for a browser based project before. I'm trying now to look at "all the possibilities", the West Wind class you suggest being one of them. I'm unsure if a web-server based solution (though still not an HTML solution) is appropriate for this app. Still pondering.

Also, I figured out how to return an array from an OLE server object! I create an instance of an "arrayObject" custom class on the client. It has a single custom array property and one method called "ReceiveArray". Then, I pass the loArrayObject to the automation server. It creates an array, ex: aServArray[12,4]). The server then calls loArrayObject.ReceiveArray(@aServArray) which copies the array to its .aArray property which can be read by the client.

Now instead of passing back delimited strings from the automation server, I can use arrays -- a big help. We'll see what develops.

Thanks again.

>Christian;
>Here's my 2 cents ...
>You may want to take a look at west-wind products. The data retrival scheme they use seems very effective. it requiers you to install a "front end" VFP app on the workstation which transfers commands to the back-end VFP application on the server (via HTTP). The result on the server side is a table which is ZIPed and transfered to the font-end app. My understanding is that west-wind product is used for the off-line reader for UT.
>
>Hope this helps.
>
>
>
>
>>I am currently exploring ways to WAN enable a VFP application that is currently under development.
>>
>>After experimenting with a VPN over a 56K frame, I have demonstrated that VFP uses "wiser" data access methods by directly opening the tables than it ever will using ODBC and remote views. For example, even if I have it set to fetch as needed and retrieve in background with a fetch size of 20 records, all 40,000 records are actually passed over the data connection BEFORE control returns to VFP. Now, at this time the cursor only has 20 records in it, and true, it loads data 20 records at a time as I scroll down a browse, but all the data has actually already been passed over the connection.
>>
>>Conversely, VFP actually handles its data in a smarter fashion. Only loads the index when a function call that uses an index is invoked. Only loads data by needed "sets". For example, if I browse a table the top (approx 20 - 40) records are read accross. If I GO BOTTOM, the bottom 20-40 are read accross, but the middle 41,900++ still have not been read in.
>>
>>Unfortunately, directly opening/accessing tables will not be an acceptable solution because it is also too slow. It would go down to "sluggish" over a T1, but it is uncertain if clients will pay for T1 lines to have a "sluggish" app.
>>
>>Questions:
>>
>>1) Is it a valid assumption that a 1.5mb line (T1) would be about 6-7 times slower than the 10BaseT LAN (discounting latency)?
>>
>>2) Does anyone know of any successfull implementations of three-tier solutions with an accent on limiting data throughput over the network?
>>
>>3) Does anyone have ANY suggestions as to implementations, approaches, etc that I should pursue.
>>
>>4) Do SQL Server ODBC drivers actually work "as expected" with remote views (unlike the VFP ODBC driver)? I am willing to consider putting the data store into SQL Server for an "enterprise" edition of the app if necessary (read "aarrrrgghhh").
>>
>>A web interface is not really an option due to the complexity of the data involved. It would REALLY SUCK under this implementation.
>>
>>I am interested in referrals to reading material, technical papers, new ideas, and perhaps best of all, a three-tier framework if one exists.
>>
>>Thanks very much in advance,