Re: [tree-ssa] Java: convert byte compiler to functions-as-trees

[Andrew Haley <aph at redhat dot com>]

Jeff Sturm writes:
 > On Wed, 23 Jul 2003, Andrew Haley wrote:
 > > Once we do this, we can beging to do tree optimizations in the byte
 > > compiler.  Such optimizations include inlining a bytecode method in
 > > a source compiled method.
 > 
 > Yay!  I'll try this in my tree tonight.
 > 
 > There'll need to be a way to inline but not expand external method
 > references.  A little like C++ "extern inline".  Have you thought about
 > how that will work?

Nope.  Not even a little bit.  :-)

I'm not sure I even understand the problem.

 > > Unlike the source compiler, the bytecode
 > > compiler generates BIND_EXPRs directly, which somewhat simplifies the
 > > handling of debug information.
 > 
 > I struggled with gcj BLOCK nodes while working on the mainline cgraph
 > patch.  They create no end of confusion.  What do you think about getting
 > rid of them altogether?

I'd approve that in a blink.

Andrew.