Hi, > -----Original Message----- > From: Bruno Dumon [mailto:[EMAIL PROTECTED]] > Sent: Wednesday, November 27, 2002 1:25 PM
> On Wed, 2002-11-27 at 22:10, Robert Koberg wrote: > > Hi, > > > > > -----Original Message----- > > > From: Bruno Dumon [mailto:[EMAIL PROTECTED]] > > > Sent: Wednesday, November 27, 2002 1:04 PM > > > > > > It just means if you use Saxon, Saxon will use Aelfred (which is > > > bundled in the > > > > Saxon jar) by default. > > > > > > > > If you want Saxon to use Xerces because you want a more feature > > > rich parser, you > > > > need to set the system property to point to Xerces. > > > > > > > > But, it is not optimal. Saxon would work much faster using its > own DOM impl. > > > > > > Saxon can only work with its own dom implementation. Even if you set > > > xerces as parser, saxon will still build its own optimized "dom" tree > > > (but will use xerces as sax parser). > > > > Ok, sorry, but it is still a much slower thing because of this. > > > > Nothing to sorry about :-) But why would it still be much slower? The > parser is only used for parsing the stylesheet, after which the > stylesheet would normally be cached in a compiled state. The little > difference in the one-time parsing speed of the stylesheet shouldn't > make any important difference. > I have been searching for threads from Michael Kay on this subject, but saxon's sourceforge archives are not searchable. Basically, I was just parrotting what he says about the subject (I think - it has been a while). What he does, and recommends is what I do now. Instead of incurring the overhead of the DOM (even though the nodesets I created were pretty small) I just set up a string in the session and use an URIResolver to return a StreamSource(StringReader(str)) to a xsl document() call - simple, and no DOM. best, -Rob URIResolver evangelist :) --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
