Boban,
>I am suprised that this totally kills VFP, I think that isn't to much for processing XML.
It's not killing VFP, it's the XMLDOM that's the problem. It's too much data for the XMLDOM to handle. It is constructing the document in memory and it's size greatly increases. And as I said before it's not a linear function of document size.
If it's a simple XML you can write your own simple string based parser in VFP pretty easily. The new strextract() function makes this pretty trivial, and it would be a lot faster then the XMLDOM.
You might also look at something like SAX.
>There are 4 columns. In two there are 50k rows, and other two are empty.
>Yes, CPU usage is 100% but usage of memory doesn't increases. In this case there are 100 Mb available memory.
>I have tried also with 5 MB (10.000 rows) document but it's same as with 16 MB.
>Any solution?