Re: .class code optimization

[Andrew Haley <aph at redhat dot com>]

Tom Tromey writes:
 > >>>>> "Adam" == Adam Megacz <gcj@lists.megacz.com> writes:
 > 
 > Andrew> Because, as far as we are concerned, there's no point.
 > 
 > Adam> Inlining methods within and across .class files makes a huge,
 > Adam> huge, huge difference if you're deploying to the Microsoft JVM.
 > 
 > It is probably easier to write a new tool that optimizes the .class
 > files directly than it would be to teach gcj how to do optimizations
 > before generating .class files.  This might even be true once the new
 > tree optimization code is complete; it is possible that SIMPLE won't
 > retain enough information, or will do too many transformations, for us
 > to generate bytecode after lowering.

It's certain IMO.  SIMPLE is far too low-level to generate bytecode.

Andrew.