>>Well, structure is data as well - I'd probably view the XML as a type of protocol, but implement creating the payload via "structure extended" or afields().
>>Creating the the cursor client side would be reading in the field descriptors in a standardized table or array again and use vfp commands ;-)
>>
>>That way server and vfp clients have the benefit of speed, brevity and clarity while still keeping the transmission protocol "internet typical&friendly"
>>
>
>I assumed Dmitry was not authoring the cursor structure by himself, but instead getting it via Web Services in an XML document, under the urn:schemas-microsoft-com:xml-msdata namespace. It is simpler than that, now I know, and XML encoding seems that would just overload the process...


Well, I am often thinking/working from the unsaid assumption that a DB developer works from a Data Dictionary, even if MS has been pushing code first in Dotnet for a while and in NoSQL DD is often an afterthought, if happening at all outside some hints in the documentation (naaah, I am not old-fashioned...). From that approach generation of structure details flows naturally to piping out afields or structure extended info - also not really on mark in this case ;-)