From: Godmar Back (gback@cs.utah.edu)
Date: Wed Dec 09 1998 - 17:10:49 EST
Also, needless to say, we still must support pizza and/or javac because
we want to run on systems that do not allow the execution of native code.
At least, I want to.
Unless you're talking about *linking* it with kaffe (that would actually
be a cool option) --- java/javac combined:
if (!strcmp(argv[0], "javac"))
dojikes();
else
if (!strcmp(argv[0], "java"))
dokaffe();
One thing that would also concern me is how optimizing jikes is?
Is it? I know that javac -O produces *a lot* better code than pizza.
I wonder how good Sun's Java2 javac is.
- Godmar
>
>
> Since slashdot has already announced that kaffe+jikes is a complete
> open source solution, we might as well make it true...
>
> My concerns: first, could you try Edouard's LostFrame test and see
> whether kaffe is in fact unable to handle jikes-generated code?
>
> Second, what about the bugs in Jikes that prevent the run-time library
> from compiling? If you fix them, will the result still be compilable
> with javac/pizza?
>
> - Godmar
>
> >
> > I've just compiler Jikes and run it over my class files (I've a number
> > of fixes I'll submit today so it works). It looks really good (took
> > 2.89 seconds to compile the whole class library on my 450MHz
> > Pentium II). How would people feel about moving over to it since it's
> > Open Source (it is right?) and Pizza isn't? The downside is that it's
> > in C++ rather than Java so it's an extra bit of compilation.
> >
> > Regards
> > Tim
> >
> > --
> > Tim Wilkinson Tel: +1 510 704 1660
> > Transvirtual Technologies, Inc., Fax: +1 510 704 1893
> > Berkeley, CA, USA. Email: tim@transvirtual.com
> >
> >
> >
> >
>
>
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:09 EDT