Just as it is feasible to statically link bytecode to a kernel, it is just
as feasible to statically link machine code to a kernel. Let me explain.

I'm thinking specifically of boot packages: java.lang, java.util, java.io
and java.net.

In order to create a bytecode resource, bytecode must be translated into a
resource. Translation is required. My translation program reads bytecode,
creates C++ source code, compiles it and outputs an object file (.o or
.obj). The bytecode is stored byte-for-byte in a rc_Bytecode wrapper class
inside the object file.

.class --> .o

On the other hand, a programmer could write C++ source code for each boot
class, compile it and output an object file. Or, a translation program
could read bytecode and output equivalent machine code. Pre-compiled for
boot classes could be statically linked to a kernel. Like a kernel bytecode
cache, a kernel could provide a "cache" of machine code for virtual
machines to share.


_______________________________________________
Kernel maillist  -  [EMAIL PROTECTED]
http://jos.org/mailman/listinfo/kernel

Reply via email to