[Jscheme-user] Re: Fwd: Re: Best Java Scheme implementation?

[Ken A. <kan...@bb...>]

This is an interesting assessment.  SISC is a very nice Scheme implementation.

Anton, could you say more about what your liked about SISC's generics over java dot notation?  In the first version of JScheme (SILK) you had to mention each method you wanted to use from a Java class and that became pretty tedious quickly.

I looked at 2 of th SRFI implementation and there looked like there were about 30 lines related to classes and methods and type conversion that wouldn't be needed in JScheme.  That might be few enough to live with.

It doesn't look like SISC much with its Java side.  For example, there isn't a gui library like JLIB, or Tim's collaboration stuff.

I think JScheme should stop advertising itself as R4RS, since its mostly R5RS.
It was certainly true that JScheme may have looked dead since its downloadable jar hadn't changed for a long time.

k

At 04:47 AM 9/24/2003 -0400, Geoffrey S. Knauth wrote:
>Some JScheme feedback I saw from someone whose 



>opinions I respect.
>
>Geoffrey
>
>------- Begin Forwarded Message -------
>
>Subject:     Re: Best Java Scheme implementation?
>From:        Anton van Straaten <an...@ap...>
>Newsgroups:  comp.lang.scheme
>Date:        Wed, 24 Sep 2003 04:32:37 GMT
>
>Russell Wallace wrote:
>> I'm looking for a Scheme that targets the Java platform. I might try
>> out two or three of the available ones for myself, but looking for
>> pointers first as to which are the best ones for a short list? My main
>> criteria are reliability, reliability and reliability, with
>> reliability coming a close fourth; I'm also looking for arbitrary
>> precision rational numbers, low-overhead compiling at run time, and
>> ease of integration with Java code (though I imagine the latter can be
>> accomplished by writing wrapper functions in any event).
>
>I've been using SISC when I need to target the Java platform, and have been
>very happy with it.
>
>I previously tried JScheme, initially attracted by its syntax for accessing
>Java objects.  However, JScheme is mostly R4RS, not R5RS.  Neither
>syntax-rules nor syntax-case are provided as standard, although it does have
>define-macro.  Another point which mattered to me is that JScheme's
>continuations are restricted to being escape procedures, but I like being
>able to use full continuations in web applications.  Finally, I have the
>impression that JScheme is the least actively developed of the Java Scheme
>implementations.
>
>SISC supports syntax-rules, syntax-case, and full continuations.  In
>addition, its Java integration turned out to be excellent (which I didn't
>fully realize at first glance) - it uses generic functions to allow Java
>methods to be called like ordinary functions.  In the end, this actually
>seemed cleaner to me than JScheme's "Javadot" notation.
>
>SISC is an interpreter, but it's a sophisticated one which performs very
>well.  The paper on its implementation is worth a look:
>http://sisc.sourceforge.net/sisc.pdf
>
>Kawa is also a very solid implementation, which I've worked with just a
>little.  It has a document comparable to the above, at
>http://www.gnu.org/software/kawa/internals.html (I find that info about
>internals helps me make choices.)  Kawa compiles to bytecode, which may be
>an advantage, depending on your requirements.  However, this also results in
>some restrictions related to continuations.
>
>I have no experience with Bigloo, but I believe it's fairly unique in
>compiling to either C, or Java bytecode.  Its C-targeting is supposed to be
>very good, and I would imagine that the Java is good too.
>
>Anton
>
>-------- End Forwarded Message --------