I think DBC/DBFs are the weakest point of VFP - otherwise an extremely strong, datacentric oop development environment (with the obvious limitation of not being part of MS's future strategy ) 

I've been doing all my VFP development against SQL backends for the last 5 years or so and I believe that is the way to go for current VFP developers and especially for learning Strataframe and .NET.  The OLEDB driver for VFP is, IMO, still very flakey - especially if you need DTS or SSIS and there is really no up-side to using a VFP data store, now that SQL Express 2005 is available.   Converting VFP tables to SQL data really isn't a big deal (especially if using a tool like FullContact and Strataframe's Data Deployment Toolkit - a combination I've been working with for a year that has really taken the grunt work out of some very big projects)

I've done a lot of conversion of VFP to SQL data, so when I see you in Amarillo if you want to talk about some of the issues in making that conversion for both running apps and future design I'd be happy to. 

But the bottom line, IMHO - while there is much to love in VFP (and the eTechnologia stuff may allow us to leverage some of that for years to come) the data store isn't part of that.  There really is much to love in SQL Server and it is easy to see why folks like Gary DeWitt and Jim Duffy were telling us 10 years ago it was time to move in that direction.