Brent Worden wrote: > > If you're talking about creating a Service Provider Interface, I > whole-heartedly agree. I envision we would follow the existing JAR > service provider mechanism >(http://java.sun.com/j2se/1.4.1/docs/guide/jar/jar.html#Service%20Provider) > and use commons-discovery to enable it. >
Yes, That is a technology we could take advantage of.
> The library, as it stands, wouldn't be too hard to morph into two > separate JARs, one for the API and one for the default implementation. > Brent Worden > http://www.brent.worden.org
The questions for me that arise have to do with applicability to other libraries like JLink, Omegahat, R, S-plus, Matlab etc. In an ideal world, it would be an ambitious project that would require involvement from all these groups to establish a standard API (in the real world, estabilishing an example of such an API is probibly more attainable).
I'm surprised that nothing of this nature has already developed? Unless I'm missing something in my research. There is an interesting project at Omegahat for a standard StatDataXml markup language for transfering data between platforms. But I have yet to locate anything suggesting a a standard API of some sort for Java itself.
http://www.omegahat.org/StatDataML/
And Omegahat itself is a platform that has several plugins spanning many of these applications.
http://www.omegahat.org/download/R/packages/
So in summary, so far I have the following examples for which such a standardized API might be usefull:
SJava: JNI bridge between Java and R. http://www.omegahat.org/download/R/packages/SJava.tar.gz
Matlab has their Bridge between Java and Matlab http://www.mathworks.com/access/helpdesk_r12p1/help/techdoc/matlab_external/ch_java.shtml#82500
Mathematica has the following JLink API http://www.wolfram.com/solutions/mathlink/jlink/
Finally, CERN has the COLT package http://hoschek.home.cern.ch/hoschek/colt/
I would find it interesting to review these API's to see their similarities and differences in relation to each other and Commons Math.
-Mark
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
