cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Berin Loritsch <blorit...@apache.org>
Subject Re: [provocative] resurrecting native code
Date Fri, 15 Feb 2002 21:18:22 GMT
Stefano Mazzocchi wrote:
> I've (finally, some would say) come to the conclusion that WORA (write
> once run everywhere) has to do with Java more or less like it has to do
> with any other programming language in the world. 
> 
> Despite Sun's marketing.
> 
> Thus, we (Pier and I) have decided break the unwritten rule "don't mix
> java bytecode with native code" and decided to go resurrect native code
> and use JNI.
> 
> Early investigations are *impressive*. 
> 
> I even venture to say that the right mix of java code and native code
> could well outperform completely native implementations.
> 
> This said, I want to throw a stone in the lake and see where the waves
> go:
> 
> if Cocoon performance bottleneck is XSLT processing, what about using
> Xalan C as the XSLT processor instead of Xalan J?


Couple of questions:

1) isn't the bottleneck of method invocations between JNI and java code
    much higher than simple java invocations?  SAX can have *ALOT* of method
    invocations, so this can be a major issue.

2) according to the last published benchmarks, Xalan C is slower than Xalan J
    (http://www.datapower.com/XSLTMark/).

3) as much as I hate to say it, M$ XSLT transformer outpaces everything
    out there according to the afformentioned benchmarking.



-- 

"They that give up essential liberty to obtain a little temporary safety
  deserve neither liberty nor safety."
                 - Benjamin Franklin


---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org


Mime
View raw message