Re: Porting the MCL Compiler to Powerpc?

[bill@cambridge.apple.com (Bill St. Clair)]

At  7:08 PM 3/7/94 +0000, olivier.b.clarisse wrote:
>What is so difficult about porting the MCL compiler
>to another hardware chip?
>There are companies specialized in binary ports, I have
>met one of them (Echo Logic). The principle is that
>all CPU's look alike at some level and there exist
>tools that can automatically translate programs from
>one binary to the other (e.g. 680XX Mac and SPARC).
>It can be easier to translate from binary to binary... Why not MCL?
>Even the Powerpc is only a CPU after all!
>
>Apple knows about binary translation, I am sure it's
>already been used, so why not apply it to MCL?

Remember, we're talking about a compiler here. It spits out
68000 machine code. Sure, much of MCL could be made to work
by a translator (closures and CLOS generic functions would
not work since 68000 machine code is part of the functions
that create them. Also, MCL's disk image contains compressed
code in its data fork.), but the compiler back-end needs a rewrite
that no translator I know of could do.