Hi, Carlos

>I have a software created with visual studio 2003. This is complete set of
>webservices, to control my inventory software. The functions return a
>datasets or XMLDocs. I´m trying to consume this datasets or XMLDocs in VFP
>8.0. My problem is when I have more 2000 records, the curfill function takes
>a long of time to fillout the cursor, I´ve been test the function, and when
>1000 records the cursorfill takes about 0.8 seconds, with 2000 records about
>2.8 seconds, but with 3000 records takes about 10 to 12 seconds, with 5000
>records takes about 60 seconds, as you can see it´s exponencial.
>
>My questions are:
>1. What is the better way to recover a .net dataset from VFP?.
>2.What is the maximun size that supported by cursorfill in XML?.
>3. An XMLAdapter recover the dataset faster than a cursorfill method?
>4. Is there a library or something like that to covert xmltocursor in
>a faster way than the original VFP functions.?

The problem you have with XML is that the MSXML parser needs to load the entire XML document into memory before converting it to a cursor. Of course, once you run out of physical memory, you start using virtual (swapping to disk), and teh performance goes dramtically down.

As discussed some time ago in another thread, you can process this a bit faster manipulating the XML parser by hand. If the record volume can grow even bigger, then consider using SAX to parse it, as it allows you to do it without having to read the whole document, but element by element.

You can also do some tests with Rick's wwXML class (available at www.West-Wind.com).

Hope this helps,