Hi Jeff, > > Thanks Huib, > I'm afraid my document is very interconnected. I'm coming to realize that > xslt has pretty much a hard limit on the size of files you can process with > it, i.e. if it can't fit into memory and you need the full power of xslt > you're pretty much out of luck. What about going to a 64 bit machine and > just allocating a huge amount of ram to Tomcat? Can this been done with > tomcat/cocoon?
It works until XSLT needs more than the available RAM. I'd try STX (Joost) first, it works very well. The interconnectedness of your document might spoil it though. You have another option: store the XML in a database (BaseX or eXist for instance) and use XQuery. My guess is that could improve things as I imagine it is not necessary to have the entire document in memory since it is already addressable 'in the database'. YMMV. And then there's "The XML technology that they don't want you to know about" option (http://vtd-xml.sourceforge.net/). I haven't used it, perhaps because I didn't want to know about it ;-), but it is claimed to be able to parse and process (using XPath) 256GB documents. Huib. > > Van: "Steven D. Majewski" <[email protected]> > > [1] You might look at using Joost/STX which does streaming transforms. > It's not XSLT -- it's an xslt-like transformation language designed for one > pass processing, > so STXPath is more restricted than XPath. > > http://joost.sourceforge.net/
