On Fri, 2003-06-27 at 18:36, Charles Duffy wrote:
> On Fri, 2003-06-27 at 11:15, Dan Polansky wrote:
> > What seems to speak for Python is that it claims to be
> > intended for general public. What seems to speak for
> > Java is that it is probably one of the standards of the
> > future, and that I like it :).
>
> A comment here: Jython compiles Python code into Java bytecode, permits
> use of Java runtime library classes, and can be used to create objects
> which can be used in Java code.
>
> If you consider Java-the-bytecode the standard you speak of, therefore,
> Jython is quite compliant. Personally, I think this -- having multiple
> languages compiling to a single portable bytecode -- is a Very Good
> Thing; Python, for instance, is far more suitable for rapid application
> development than Java, whereas Java is more suitable for very large,
> multi-programmer projects. Being able to interoperate provides the best
> of both worlds -- allows, for instance, a quick script to be developed
> in no time flat that interacts with the persistance layer of a large,
> multi-programmer project.
There is a framework that makes it possible to embed several different
scripting tools. BSF http://www-124.ibm.com/developerworks/projects/bsf
it's a IBM Alphaworks so I don't know about licenses, have not had time
to look at it. But using the BSF makes it possible to support different
scripting systems. As long as they comply to the BSF interfaces. More
information can be found at the mentioned web page.
>
> Similarly, for permitting an easy way for folks who aren't familiar with
> Freemind's internals, a mechanism for extensions via Jython would be
> useful.
>
There are alternatives to Jython as one BeanShell
(http://www.beanshell.org/) that is a the author puts it 'Lightweight
Scripting for Java' that supports BSF.
> ...but wait! Because Jython uses standard Java objects and interfaces,
> and its objects appear as standard Java objects, a mechanism written to
> support extensions in Java could be used by folks doing Jython
> extensions as well! (ideally with a touch of glue code and some
> forethought during design to make the interfaces more conformant with
> The Python Way). So we can have it both ways -- easier extensibility via
> Jython *and* Java.
>
Yes this could be the long term plan, as we need to get the interfaces
to the nodes and the entire node model in place before we can start to
communicate outside the core of Freemind. Dan and Joerg do you have any
good suggestions about where to start?
Would be nice if someone would participate in working out a preliminary
interface description to the node model that would enable us to test run
the different interpreters and have a better look at the features.
... I'm starting to wonder how many private 0.6 source trees are out
there? With high local customizations, that cannot be synced easily. I
know that I have one heavy modified one. But this is most for study of
the flexibility of integration of interpreters or use of native classes
loaded on demand.
/
Peter
|
