[XML-SIG] 4XSLT focus on docbook stylesheets (and fruit thereof)

[Uche Ogbuji]

> [Uche Ogbuji]
> > Which is noce, but I knew I could do better.  Turns out that was
> > understating it.  After the big move: making key computation lazy until
> > a key is requested in each doc's context, we go all the way down to:
> >
> > time for 10 runs: 17.4807579517 seconds
> > time per run 1.74807579517 seconds
> >
> >
> 
> Outstanding!  Any ideas how that compares with the big guns (Xalan, msxml,
> saxon)?

I can't do it all on my own.

I'll be happy for contributed (and careful) benchmarks, so that we can 
document to users what to expect.

However, I must say that I don't expect to do any more major optimization 
passes on the current 4XSLT.  Given its fundamental architecture, we're 
approaching the point of diminishing returns.

We have plans for a complete re-architecture in 4Suite post 1.0 (which I think 
is actually in sight now), after 4XPath and 4XSLT have been moved into PyXML.  
These are quite mature, conformant, have extremely strong Python integration, 
and after recent work by Mike Brown, Jeremy Kloth and me, have respectable 
performance, although I don't expect it to outperform msxml or libxslt for 
obvious reasons.  I think this makes them great cantidates for the PyXML 
built-ins.

We have ideas that would make the next generation compete with the C XSLT 
processors, but that's a different story.


-- 
Uche Ogbuji                                    Fourthought, Inc.
http://uche.ogbuji.net    http://4Suite.org    http://fourthought.com
Track chair, XML/Web Services One Boston: http://www.xmlconference.com/
The many heads of XML modeling - http://adtmag.com/article.asp?id=6393
Will XML live up to its promise? - http://www-106.ibm.com/developerworks/xml/li
brary/x-think11.html