Thanks Christian,

So what about sending the data to memory, as they're doing, then bring it to a VFP array using your library, make the modifications in the array with VFP, and then sending the array back in just one time ?

That would avoid all those SYS(2600) calls...

Is that correct ?


>Hello Cesar,
>
>"What about those Marshaling functions that you provide ?
>How fast are they ?"
>
>They are quite fast, especially the marshaling functions that convert VFP arrays/cursors to C arrays are several hundred times faster than anything one could write with VFP code.
>
>But how does this relate to this topic?
>
>These functions were designed to emulate C structures easily (semi automatic conversion is done by the provided struct translator application) from VFP without thinking about datatype conversion, structure alignment etc. .
>
>VFP is just the wrong tool for any algorithm that has to deal with data on a per byte basis.
>
>Christian