On Friday 04 June 2004 14:29, Jeff Kenton wrote:
> There are people who have 100 megabyte product catalogs in XML files.  They
> obviously don't serve the whole thing to a web page, but they do use XSLT
> to process those catalogs, and produce small web pages or XML output for
> reports.
>

We have seen really two types of request coming for very large transform 
support. The first is from companies where the data size is within the range 
of an object model based implementation, but who are aware that the ceiling 
for such solutions is not far above them. The second is companies who are 
doing data streaming (with embedded transforms) between systems. While you 
could argue that the first group will sleep better if they employ 64-bit 
machines this really does not address the CPU costs issues. The second group 
is harder to deal with in that they can view XSLT as just another pipeline 
process. To be fair there are not a significant number of companies with 
these problems but they do exist.

Kev.