Industrial-strength (non-Java) XSLT app for OS X?

[ĸupiэȶƺ]

Frustrated XML/XSLT newbie here with a question - Can anyone recommend a powerful, non-java XML/XSLT program for OS X? I've been working on transforming FileMaker Developer's DDR report XML files, which are up to 25MB in size.

I've tried the popular java XML processors (Oxygen, Editix) and they are much too slow to use... when trying to open the huge DDR output files, I gave up and force-quit Oxygen after 13 hours (!!) of the "beach ball" wait cursor. Editix simply tries for a while then says it can't open the file.

I've also been using the public domain "TestXSLT", which sometimes works but generally behaves too much like public domain software for me to get anything done with it: too many unexpected quits, features mysteriously refusing to function, it sometimes inserts garbage characters into the XML, etc. I also was told Smile had some XSLT features, but it's so complicated and technical to me it might as well be written in Latin.

I know none of this is my XSLT, because when TestXSLT is in the mood to work properly, the transformation works fine, and takes under 90 seconds.

Anyone other Mac users have better suggestions? What program are you using to write/troubleshoot your XSLT on?

Thanks,

Mike

[Garry Claridge]

This type of thing could be written in RealBasic without much bother. I guess that if enough demand exists this would be a worthwhile project. Pity I don't have much time until early March

Good Luck.

Garry

[Martin Brändle]

You might try the C++ version of Apache Xalan-C and the Xerces-C XML Parser, however, there is no OS X binary for Xalan-C and you have first to port the code and to compile it. When you have the binary I would be interested

I have noticed your message also on the XML_Talk list. Did you follow the Smile path further?

If you have a Windows computer, another way is Michael Kay's Saxon that you can call from the command line. I had to use it once to transform a 45 MB XML output, because with the XLST engine built into XMLSpy this was not possible due to the memory requirements (the builtin engine blew up to over 2 GB). With Saxon, the transformation required about 600 MB of memory and went through in 1-2 minutes. The problem is that an image of the XML tree first must be built in memory, before it can be transformed.

Martin

[cjaeger]

the xalan-c/xerxes-c parser is built into FileMaker. I sometimes export a dummy record from FileMaker, using a XSLT which calls another import doc using the document() function.

Works great (after debugging the xslt in "TestXSLT").

[Martin Brändle]

Good trick, Christian! I referred to FMS7A probably having the Java versions of Xalan and Xerces, but did not know of the C versions in FM7 Pro.

Martin