Cyril Hansen wrote:
> My problem is that, at the moment, Gcj assumes the complete application
> is compiled and other members in our group are planning to use advanced
> JVM features like dynamic class loading, beans, introspection...
>
> So far I think we need to interface the compiled Java code with a
> real JVM. Some old Gcj documentation says they have planned to use
> Kaffe for this purpose, but I see nothing in the code.
I suppose that you will not be deploying it in final form anytime soon.
In that case I think you can safely suppose that some state-of-the-art
jit compiler will appear for linux JDK - be it hotspot, IBM jdk or
anything else. It is possibly even faster for long running programs and
program is a lot more maintainable.
And for now stick with JDK + possibly tya - I don't think that chasing
bugs in japhar or gcj is good at early stages of project :)
Artur
