I implemented the certificates via winHTTP and in essence wrote a small XML-RPC handler I partially re-used for web services.

Fair enough. In FPW2.6 it was accepted practice to write all sorts of stuff in FP- even visual elements. I guess that was "the norm" back then when IT was a frontier, not a commodity. Now we have little choice but to use others' complicated black boxes with all their idiosyncrasies and shrouded bugs. ;-)

Yeah, web services can be a pita. IME just about any sort of message is easy enough to generate with VFP's textmerge, also easy enough to read IME if you chrtran() the XML tags and import the whole thing into a VFP cursor and process from there. I agree this is one area where vendor black boxes can create a feast of trouble, especially if vendors want their own version of the "standard" to prevail ;-)

The bad behavior sounds worrying. I've not seen this sort of flakiness, though I did see heavy SQL drag the machine to a halt in 1995-1996 (similar to a large ADO.NET recordset ;-) ) That time it was a memory leak in VFP. Who knows, perhaps there's a sneaky bug that reveals itself only when you combine a certain sequence of activities without the need to garbage-keep. I guess that's one of the weaknesses of VFP's version stability- improvements are piled upon stuff that may be considered "stable" but which makes assumptions that may be no longer true. That's one of the main potential benefits of the Etecnologia efforts IMHO: the worst of the legacy is swept away while functionality remains.