RE: XSLT on Wikipedia

> You'd probably want to sax parse the 14gb file :0) and then 
> for each <page> element run the upconversion and then sax 
> parse the result of that, sending all the events to the same 
> handler... or of course just write out each page into a seperate file.

If each page can be processed independendly then it's probably very amenable
to the "streaming mode" processing available in Saxon-SA:

http://www.saxonica.com/documentation/sourcedocs/serial.html

Michael Kay
http://www.saxonica.com/