RE: [Sisc-users] SISC vs Kawa
Brought to you by:
mradestock,
scgmille
From: Dominique B. <dom...@nu...> - 2005-11-15 23:35:16
|
> SISC will in general be slower than Kawa, though it can be faster, > and for certain things (like continuations, which combined with > serialization seem to be the hot place for using SISC), it is more > than an order of magnitude faster. Its important to test > the functionality you'll actually be using. I second that. I have chosen Kawa over SISC strictly for performance reasons based on a benchmark for some typical code in my application domain. But YMMV based on YOUR application domain. And confess that I sometimes miss TCO and true continuations... So it's always a matter of compromises. > SISC is more complete in more ways than just standards > compliance. It > also has a much more complete standard library, far more SRFI > support. > It tries to be a complete Scheme environment, rather than > a way to write Scheme code which is compilable as JVM bytecodes. For > that reason, Kawa relies much more on one calling into Java libraries > than SISC. I agree with Scott. Many more SRFIs are available with SISC when compared to Kawa. Another interesting aspect of Kawa is the ability to use a Java debugger to single-step the code, inspect local variables, etc. This is possible, for example, in Eclipse with the SchemeScript plugin (http://schemeway.sourceforge.net), which offers some support for Kawa. Dominique Boucher |