Re: MATLAB-like Tool

I think the idea of any scripting language is to prototype and interact and
try things.  A big system, where speed matters, should be built with a
statically typed language like Java.  So I don't think speed is an issue for
scripting languages.  BeanShell will always be slow because it doesn't
generate bytecode.  Pnuts is probably the best Java scripting language, and
it does generate bytecode, but it has a tiny user base and an author who is
almost as unresponsive as BeanShell's author.  I picked BeanShell as a
compromise because it has a fair size user base, and while not as good as
Pnuts, is not nearly as bad (for me) as Groovy.  I was hoping that
BeanShell's user base would take some initiative if I forked the source and
made it easy to contribute, but unfortunately, this hasn't happened.


Bob Fischer wrote:
> 
> BeanShell is known to be pretty slow in general, so  
> that's a big reason NOT to use it for this purpose; one would start  
> building up some routines and then discover that the interpreter is  
> not fast enough to run time.  Then you're kind of stuck.
> 

No, then you translate your time-critical routines into Java.  This should
be painless compared to, say, translating code from Python to C.



> The PySci  
> approach looks promising; I would want to see that it integrates well  
> with Java when running under Jython.
> 

This won't work because SciPy is based on a big C library and Jython is pure
Java.

-- 
View this message in context: http://www.nabble.com/Hacking-BeanShell-with-JavaCC-tp16715411p16753214.html
Sent from the BeanShell - Dev mailing list archive at Nabble.com.


-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone