Rich,

Thanks a lot, the experience you are sharing is a great eye opener.

A question, then, what would be your approach if you would begin doing it all over again?

Max.


>Maximo,
>
>FWIW: I'm nearing the end of a conversion from a 2.6 DOS program.
>
>This was supposed to be a "middle of the road" conversion, between "quick and dirty" (get it running in VFP with as few changes as possible to the DOS code) and a full-blown ground-up redesign.
>
>If I can help it, I'll never take this "middle of the road" approach again. The idea was make the UI pretty, but keep anything that's "mostly processing" as unchanged as possible. So now we have a VFP menu and some VFP forms, and a bunch of calls to old, mostly unchanged prgs.
>
>Problems:
>
>1) The old system had very poorly designed, non-normalized data structures, which we were required to keep, in order to keep as much as possible of the "procedures" unchanged. This includes a separate set of tables for each (user) company. So you can't put tables in a form's data environment, because you open different tables depending on a global memvar setting when you call the form.
>
>2) The old programs make many assumptions about private variables remaining in scope, when in the VFP version they have gone out of scope.
>
>3) Some of the prgs are still doing some @/SAY/GET/READ stuff (on a "blank" VFP form instantiated early in the new startup prg, and activated whenever some old DOS screen code will be run). So the interface isn't consistently Windows-like.
>
>Making this work has probably taken almost as long as it would have to do a ground-up redesign, and the result I'm about to deliver is something I'm not very proud of.
>
>Enough griping. As I said, I'm near the end of the conversion.
>
>Regards,