As a developer with experience dating back to the FoxPlus and dbase II days, and located in Texas, I tend to agree with Michael.
As for the days of FoxPro being numbered, I tend to agree there too, but maybe for different reasons. Microsoft is trying to position FoxPro as NTier, while continuing to advance Access and Office Automation as the way to go. I develop for the defense dept, and am not a consultant, but lead a team of full time VFP developers.
As for training, my personal survey conducted un July 2000, revealed NO colleges/universities in the US were offering training in FoxPro of any version or flavor, and most of them offered training in Access, and/or Office. The only college level courses available in VFP are located in India and Malaysia, by universities that Microsoft owns. Condensed training is offered ONLY by Jim Booth's outfit, and while it is good, the condensed courses, allowing for transportation, etc. are really too expensive for many of the beginners. There are NO certification guides available as there are for most every other programming environment from Microsoft. One is planned for release early next year covering VFP 6.0/ but it will be too little too late, and VFP 7 is scheduled for realease at the same time. Last year's beta certification exams unfortunately catered to the personal preference of a few contributors rather than embrace the functionality of the product as a whole. (e.g. 55% of the exam on stand alone apps were confined to COM, and even more on the distributed exam) What the result of all this ongoing lack of what I call infrastructure, is why FoxPro is becoming less known.
Many independent consultants are too quick to migrate database products to SQL/Sybase/Oracle, etc, where they are limited to communicating with SQL language. Any VFP developer worth his salt, knows that data querys using the Rushmore Optimization in the FoxPro engine far exceeds the persormance of SLQ, and SQL server. Where performance is critical, far too many "consultants" are recommending improvement in hardware and connectivity to enhance performance of an inherently slow product. Even Microsoft, with its Terradata project, installed super fast connections, fiber optics, clustered servers, etc in order to get some kind of performance out of SQL server, when that entire application could have been done in VFP, run on an NT server, and provide query results just as fast as all that expensive hardware (In my opinion)
Companies that do have development staffs, usually don't know how to measure performance well enough to go with a consultant, or outsource portion of application development. Unfortunately, this is a gap the consultants have not yet filled, that is the education of the client as to how to measure outsourced performance. Remember that when doing a query in VFP against a table containing a memo field will fetch the memo contents and preserve its formatting? Try that with JDBC, or any of the other front line implementations of SQL. It just won't work.
After all, some very large corporations are dealing with flat file databases (ascii text files)to this day and they still work.

I still love FoxPro, and I am happy the DoD still allows us to demonstrate the performance capabilities of VFP, although it is getting much more difficult to recruit young blood into the business with the lack of "infrastructure." The other problem is that DoD contractors cannot readily outsource to a consultant, due to security requirements.

Just my current rant.