---- you Morten Jorgensen <[EMAIL PROTECTED]> wrote ---- > Glenn (and everybody else wanting to use XSLTC with TrAX), >... > internal classes that are supposed to be plugged in behind TrAX. > The whole idea behind the JAXP/TrAX API is that you use the defined > API only, and that the only thing that links your code to our > implementation is the "javax.xml.transform.TransformerFactory" > property. Hear, hear! I strongly echo this sentiment, for xsltc and xalan users both. The JAXP set of API's (javax.xml.*) are very well designed, and even won an award at JavaOne earlier this year. Actually, I'm realizing that we should have provided even more specific examples for our Xalan-J 1.x users to convert old-style api's to JAXP. Using them makes you processor-independent, and they mirror the set of parsing APIs that also make you processor independent. In fact this project is already seeing the benefit of this: the great majority of our tests run unchanged on either Crimson or Xerces, and on either xsltc or xalan - just change a system property. Long-term one of the areas that I hope we can bring together is output handling and 'specialty namespaces'. One of the features that's not obvious is xalan's xsl:output xalan:indent-amount, which would be nice to get xsltc to recognize - there are a number of little features like that that should be easy to tie together later on. (Extensions may be another question...) Oh: and of course since Morten's example uses JAXP (aka TrAX, remember), it should be equally valid for xalan mode as well. - Shane
