Re: Native Code Generation from Abstract Stack Machine

> Independently, and without knowledge of Peter Deutsch's scheme, Ian
> Piumarta did a native code compiler for Smalltalk that he wrote up in
> his thesis at Manchester University, which is called something like
> "Deferred Code generation in a Smalltalk-80 Compiler". Its a very
> similar idea.


I recently used a very similar scheme to translate Java bytecodes to
Smalltalk bytecodes, deferring the compilation of arguments until
either the end of the basic block or a method invocation. Rather than
optimization, I also needed it for correctness, to add a receiver in
front of the arguments of an invokestatic bytecode (Java has the
receiver implicit, while Smalltalk makes it explicit).


I had also used it for GNU Smalltalk's JIT compiler, which was more or
less a conversion of Ian's thesis to work on basic blocks (Ian used
an abstract syntax tree with if conditionals and while loops).


After a few years using it for several different projects, however, it
seems pretty straightforward to me, definitely not worth keeping it
secret; I doubt that there exist JIT compilers that do not reconstruct
something resembling a parse tree (also AOStA, right? :-).


Paolo

Return to the comp.compilers page.
Search the comp.compilers archives again.