Let me better explain.
I want to use XSLT not in the traditional way.
I have a hierarchical data structure implementation of my own which
theoretically can be expressed in XML. I want to convert such data structure
to another such data structure be of a different hierarchy.
I could do the following: Convert the source data structure to XML
String/DOM (creating a copy), then run XSLT, getting String of the output
XML, then parse it and fill in my data structure. This is a time consuming
process because I do several unnecessary conversion.
I want to integrate deeper into the XSLT Engine. For example I understood
that source as DTM's is the fastest, but difficult to implement, and DOM is
second best. So I can implement the DOM interface over my data structure
(not convert it but rather expose the interface).
The questions I'm left with is if I want to fill another data structure with
the output of the first, how can I do that the fastest ? I know I can go
through SAX/DOM/String but I want to know which is the closest to the XSLT
engine inner implementation so to get best performance. If for example I
will find that SAX is the most native, then I'll implement SAX interface
over my data structure (fill it from the SAX events). I also want to keep
memory footprint to a minimum - If I have to go through string then the
memory footprint will be at least as big as the string, if it is large it
will be large. If the XSLT Engine could just issue SAX without going first
through a string then I wouldn't need that memory.
Hopefully, I'm not missing anything.
Thanks,
Yaniv.
Yaniv.
> -----Original Message-----
> From: Shane Curcuru [mailto:[EMAIL PROTECTED]]
> Sent: Fri, November 23, 2001 1:14 AM
> To: [EMAIL PROTECTED]
> Subject: Re: What is the "Native" output of XSLT
>
>
> I'm not sure I understand your question. In one way, it's XML - XSLT
> is expressed in XML, and the 'rules' of XMLness are prevalent
> throughout stylelsheet processing and transformation.
>
> If you mean format of input/output as in SAX/DOM/Streams, then it all
> depends on your application. The JAXP and TrAX APIs make it very
> simple to pass in whatever Source object and get whatever
> other type of
> Result you want, in any mix. DOMs will tend to be fast once they're
> built, but will tend to take up lots of memory. SAX and Streams will
> often behave the same - fairly quick throughput and lower memory
> consumption. (Xalan effectively asks the parser to use SAX
> as input if
> you give us a Stream most of the time).
>
> - Shane
>
> =====
> <eof aka="mailto:[EMAIL PROTECTED]";
> BartSays="Nobody reads these anymore."/>
>
> __________________________________________________
> Do You Yahoo!?
> Yahoo! GeoCities - quick and easy web site hosting, just $8.95/month.
> http://geocities.yahoo.com/ps/info1
>
