>>I only use it to migrate data from VFP data to SQL databases already deployed and have never attempted to modify a SQL schema based on VFP structures. I use Database Deployment Toolkit from Microfour (Strataframe), which uses SMO, to store SQL schema metadata and modify existing SQL deployments. Unfortunately DDT is no longer a standalone product and is only available as part of Strataframe, but I hear that Red Hat's tools are very good for this with SQL server, though they use scripting rather than SMO.
>
>In a session last year in Frankfurt Rick S. touched SQL deployment and (at least at that time) felt the Microfour tool was not bugfree enough.

And at that time, certainly, Rick was correct. And that is why Microfour stopped thinking of it as a standalone for the time being and only distributes and supports it for Strataframe users and it could evolve with the product without drawing resources and attention away from core framework development.

I will say it has come a long way in the last year and is now pretty much what I want. There are still a couple of issues using UIDs and especially Sequential NewIDs but that should be worked out quite soon. It can integrate nicely into Strataframe apps making deployment of new or modified apps a lot easier.

>Red Hat was also mentioned - perhaps we will look at that again. Currently our process (based on version numbering) works well with vfp and SQL server and we have some tweaking established with the help from a Powerbuilder company we are working on a common project for Oracle. Looking to support DB2 and perhaps some free backends we check out other tools<g>.
>
>>FullContact works well for moving the VFP data to SQL, once the mapping has been tweaked. The author is pretty responsive to both questions and suggestions so it is probably best to shoot him an email about the strengths/limitations of other aspects of the product. I think there is also a full working evaluation available on the website.
>
>Already linked his site, but the upsizing works rather well (as long as vfp dates and datetimes are cleaned) because we keep it mostly down to vfp6 data types, INT PK's and only binary index from vfp9.

I always massage the VFP data with a program of my own before pointing Fullcontact at it. As you say, cleaning dates and datetimes to accept nulls, and replacing empties with nulls and where I am using guids converting any char(16) binary guids to char(36) binary and again, modifying to accept nulls (for FKs for example) and replacing blanks with nulls gives Fullcontact a pretty good idea of how to map it.