1. You'd lose the data engine of VFP. Many people here, myself included, saw that as a bad thing.
I would take the engine loss for better performing ADO.NET Provider for VFP Database. Maybe one that would perform as good as the SQL ADO.NET Provider.
When I spoke with Sam Gentile about VFP, even though he had never heard of VFP <g> he mentioned that eventually all Microsoft programming languages would either go to managed code or be dropped completely by the year 2010.
MS Office Team had push their efforts in preparation to compile Office in .NET - this was a common push within the different MS Organizations. I don't think this is the common effort for the VFP Team.
Moving VFP to managed code would be a good thing. Improving the VFP ADO.NET Provider would also be a good thing. Working for Microsoft and not paying attention to the trend is a bad thing - at least for VFP.