XSLT on really large files?

[cybersattva]

Hi everyone,

I have some pretty mongo XML source files (say 256MB or so) that I need
to run an XSL transform on. The transform works fine when the file has
only one or two records in it, but when I tried using xalan on the
whole file it ran for an hour with no perceptible results (besides a
really sluggish CPU). Has anyone else had any luck with a pre-built
XSL translator for monster files, or am I going to need to write a
program to do the translation?

Thanks in advance!

 

[Gregory Vaughan]

You might give Saxon a try (http://sourceforge.net/projects/saxon). I've
used it on XML files approaching 100MB. When running it in Java, be sure to
bump up the memory setting, as in:

java -Xmx512m

Also make sure your machine has at least 512MB, or preferably a gig or two,
of RAM. With about 40-50 XSLT rules and a 65MB input file I've run
transforms in approx 420MB under Windows XP. But if you don't have enough
real RAM your machine will constantly swap and the time will increase
greatly...