According to Roland C Dowdeswell:
>

lots of stuff about emulating other processors elided....

>This has actually been bugging me for a bit, and although I
>have not thought it through incredibly, I have been wondering
>if there is a horrible difficulty with this approach.
>

Depending on the operating system (or lack of one in the case of
certain poplar peecee program loaders...) you will have a varying
amount of difficulty...

>(You could try to take it farther, and try to port non-unix things
>this way, although I do see a lot of problems with things that
>twiddle hardware/etc.  Perhaps if you considered having a 'virtual
>machine' and how a few pieces of hardware worked, you could then
>guess what their twiddling was trying to accomplish...)
>

If you are thinking about trying to make MSDOS/WINDOZE applications
run natively by a simple translation then you are out of luck.  The
problem with these applications is that they freely mix (in unix
parlance) text and data segments.  You cannot tell what is what by any
sensible means - even trying to start at the program entry point and
following the code path will not work because if there are any
callback functions in the code you will miss them.

I suspect that a unix application would be doable though endian-ness
may be a bit of a killer.

Further reading can be found in the newsgroups that deal with
Architecture Neutral Formats (ANF).

FWIW, yes, I did have this bright idea clubbed to death on me once
upon a time ;-)

-- 
Brett Lymn, Computer Systems Administrator, AWA Defence Industries
===============================================================================
  "Upgrading your memory gives you MORE RAM!" - ad in MacWAREHOUSE catalogue.