Re: Mixing languages (Kurt Stephens)
Sat, 3 Jun 1995 10:53:55 GMT

          From comp.compilers

Related articles
Re: Mixing languages (1995-06-03)
Re: Mixing languages (1995-06-05)
Re: Mixing languages (1995-06-24)
Re: Mixing languages gbaker@rp.CSIRO.AU (1995-06-27)
Re: Mixing languages (1995-06-28)
| List of all articles for this month |

Newsgroups: comp.lang.scheme,comp.compilers
From: (Kurt Stephens)
Keywords: design, interpreter
Organization: Swiss Bank Corporation CM&T Division
References: <3pf727$>
Date: Sat, 3 Jun 1995 10:53:55 GMT
Status: RO

Richard A. O'Keefe writes
> This is exactly what Sun are doing with Java. They compile to a portable
> stack machine bytecode format, but for real performance they have a back end
> that converts that to native code. It isn't terribly hard to write a back
> end that can do a *lot* better than even a threaded interpreter, even if it
> is not quite what a highly optimising native-through-and-through compiler
> might achieve.

Why reinvent the wheel with another VM/BYTECODE system? Use a subset of
the Intel x86 instruction set and write *ONE* portable emulator for
non-Intel machines. This "bytecode" is supported in hardware by 90% of
the world's computers. ;^)
(Actually I think the x86 is a tired, old instruction set; a poor register
set, etc. but it is the most widely used and is not going away, sigh...)

I would like to see GCC packaged as a linkable library (or a server) so we
can call a function (imagine some thing like "void* GCC_compile(RTL
expression)") with a RTL tree and have it return a function pointer!

Has any one designed an embeddable, portable, retargetable "compiler"

All the best,
Kurt Stephens

Post a followup to this message

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