On 10/01/06, Svante Schubert <[EMAIL PROTECTED]> wrote: > As Saxon is developed by the only editor of XSLT 2.0 [1], a small wonder > he is ahead of all others.
I was referring to 1.0, though yes, he is currently compliant to the 2.0 WD. > XSLT 2.0 is NOT a W3C Recommendation yet, only a candidate > recommendation and this only since less than two months. > Earlier and during the time I was bundling Xalan, it was a W3C Working > Draft. As there might be ongoing chances on spec and therefore on > dependent implementations as Saxan, a usage of XSLT 2 features in > product environment are risky. Agreed, though with dev time spent on 2.0 I'd lay money on it not changing significantly at rec status. It is worth noting that there are features in 2.0 that are extremely useful though. > > A know little about his company but taking a glance on the developer > count [2] the opensource community around Saxon seems quite singular. > On the other hand I see heavy traffic on the dev mailing list of Xalan [3]. Pro's and cons both ways to my way of thinking. The source is available on sourceforge though. > > As it is not mentione yet, it is a further plus for Xalan as it has two > implementations one for Java and one for C++. Not understood. > > Concerning the comparance of features (aside of XSLT2) and performance > of XSLT processors I am in need of some more facts (preferable links). > For example, are there recent results of a common XSLT testsuite, which > guarantees the user a common feature set? For compliance, or speed or something else? XSLT has no formal (W3C) test suite to my knowledge. http://www.oasis-open.org/committees/documents.php?wg_abbrev=xslt may help. http://www.datapower.com/xmldev/xsltmark.html for performance. > And what about XML parser and XSLT processing benchmarks, which make the > parsers and XSLT processors comparable? The parser is normally external. Certainly the case with saxon and xalan AFAIK. With saxon it's settable on the cmd line or java interface. > > They should exist as usually a XML parser/processor developer should > have these for functional and performance regression testing. Ask at apache. I think both now default to their parsers. > > Please keep in mind, that a change to a different parser/processor > always generate a lot of testing efforts and comprises regression risks. Agreed. Proceed with caution is a good motto here ! regards -- Dave Pawson XSLT XSL-FO FAQ. http://www.dpawson.co.uk --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
