Re: Using Kawa to port Perl to the JVM

[Daniel Bonniot <Daniel dot Bonniot at inria dot fr>]

I am also using gnu.bytecode, gnu.expr and gnu.mapping to compile my
research language to java bytecode. I found it quite handy for this
prupose. IMHO, gnu.expr is flexible enough for many languages, using
CallExp and ad-hoc Procedure's for your specific builtins. 
To me, the most tricky part was to have special types (java scalar or
array types convertible to classes on demand) working. There is support
for that, but it's somewhat hard coded in gnu.bytecode.

Anyway, feel free to check out my source code to see how I use kawa.
Basically, my expressions are in src/bossa/syntax, and they are
translated in to gnu.expr.Expression's (straightforwardly or not).

By the way, Per: it might be a good time to update the description in
the "projects using Kawa" page. Mine has changed name, from Bossa to
Nice. It is now hosted at sourceforge, with source code accessible thru
CVS. Here is the new paragraph:


Nice is a research programming language that has a Java-like syntax, and
features multiple dispatch methods,
constrained parametric types, higher-order functions, and the new
concept of "abstract interfaces". 
The Nice compiler uses Kawa's gnu.expr and gnu.bytecode packages to
generate Java bytecode, so it cqn be of interest to compiler writers
planning to do the same.
The source code is available under the GPL licence. The project is
located at http://nice.sourceforge.net, 
where you can find binary distribution and source anonymous CVS access.



Thanks.

Daniel