xmlgraphics-fop-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andreas L Delmelle <a_l.delme...@pandora.be>
Subject Re: Performance with large Tables
Date Mon, 03 Jul 2006 17:20:15 GMT
On Jul 3, 2006, at 06:43, mcevikce wrote:

Hi,

> This is a 3rd party filter that handles CSS parsing as well.  I also
> confirmed that my XSL-FO to PDF takes 6-7 sec but my xsl-fo  
> generation is
> the problem.

I figured as much, based on the observation you wrote earlier on,  
that "most of the time is spent in the parser"...

> From a standalone test it takes 12-15 sec to create the file
> that I attached link for.
> http://www.4shared.com/file/2278690/5364bc01
>
> 3rd party library can be found at:
> http://www.re.be/css2xslfo/
> I am using the csstofopnew from this site. It is possible that the  
> filter
> might be slowing things down.It is very easy to set this up. Feel  
> free to
> try it.
>
> Are there any other free libraries that convert XHTML with CSS?

Java libraries? Not that I'm aware of, unfortunately.
Depending on the complexity of your XHTML sources, you could try to  
achieve the same effect with an XSL transformation... (AntennaHouse  
has an example HTML2XSLFO stylesheet that is referenced on our  
resources page)
In that case, if the stylesheet is well-designed, you could really  
take advantage of Saxon being the more efficient XSLT processor --in  
your current setup, the XSLT processor just performs an identity  
transformation on the readily transformed output of the filter.

Parsing and applying the CSS templates would be another matter, of  
course :/

Maybe the source of the filter is not fully optimized, and could use  
some streamlining. But then again, given the above, I can imagine  
that combining random XML input with a random CSS template is a  
demanding task, no matter how much the algorithm is optimized. I'd  
have to dive into the CSS2XSLFO source code to be able to say for  
sure...

Quickly browsing through the manual, I read that there are two ways  
of setting up the filter. Maybe you could try the second one (p.54-55  
Example 3): placing the identity transform first in the pipeline, the  
filter would then take the Transformer's output as input and pipe its  
output through to FOP ? (as opposed to right now: your filter sits  
between the XML parser and the XSL Transformer)
Could make a difference...


Cheers,

Andreas


---------------------------------------------------------------------
To unsubscribe, e-mail: fop-users-unsubscribe@xmlgraphics.apache.org
For additional commands, e-mail: fop-users-help@xmlgraphics.apache.org


Mime
View raw message