[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Tue, 26 Mar 2002 22:33:39 +1000
Erik Rozendaal wrote:
> - A JIT. Maybe based on GNU Lightning, maybe in co-operation with
> Mono, maybe something entirely new. I believe no work is being
> done yet on a JIT?
There a few experiments on my hard drive, but it will be
a few more months before they see the light of day.
The pnet runtime engine is structured to make adding a
JIT fairly easy. The bytecode verifier chomps through the
code and then calls a "coder" for each program element
that it finds. The default coder converts IL into CVM for
interpretation. Writing a JIT involves building a coder
for the native platform.
> - An IL front-end to GCC just like there is a Java bytecode
> interface. This could also serve as a (slow) JIT, especially if we
> cache the generated code.
A lot of people have talked about this, but it runs the risk
of turning gcc into a "super assembler" that would allow
people to thwart the GPL.
> - The C# compiler. If there is a good way I can work on this without
> stepping on everybody else's modifications.
Feel free to help out. I'm all over the code, but there's
always plenty still left to do. Grep for TODO.
> - The garbage collector. But I think the current (conservative?) GC
> is probably good enough for some time to come.
libgc should be sufficient for our needs. Writing a GC
that works reliably in a multi-threaded environment is
> - Thread support. Or is that already working?
There's no support for threads in the engine as yet, and no
one is working on it. There's been some work done in the
C# library, but we still need the runtime half.