Hi Mike

I have just read again chapter 7 in the book, which I had glossed over in my first reading. You can use ODBC to access VFP tables, just as you would with other languages; it is not restricted to SQL server.
There were a few references to it at Devcon, but this thread has helped to put a few thinggs in perspective. ODBC would work against you on a local machine, for sure, but once you start pulling indexes and data across a network, processing times increase by a factor of about 10, from some tests I have been doing recently.
While I understand the general scepticism of using ODBC, I think the one place it could have a significant impact is in dealing with grids. I would expect that ODBC does not use Rushmore, and therefore scans an index sequentially, skipping filtered records, but fetching only as needed, with a lot less traffic across the wire when compared to the native behaviour of VFP. I use grids quite heavily, and the performance is extremely poor, but elsewhere, performance is quite good.
I don't think I would want to use it extensively, but I am sure it's worth some development effort, particularly in the context of data objects.

Geoff



>I hope you are not taking any of this personally. Are you suggesting the SQL Server book? That book is talking about SQL Server which has to be accessed via ODBC. Please prove that VFP accessing data via ODBC will outperform VFP accessing data without ODBC.
>