Ken writes:

> Gavin replied:
>
>  > Nahhh.  Nothing needs to be this complex.  We would
>  > simply give you the ability to make procedure calls
>  > from the emulator environment into the outside host OS
>  > environment (Linux, say), and also give you a library
>  > from Linux that could load and call procedures inside
>  > the virtual 3000.  This one facility solves all your
>  > problems in this area.
>
>  Who knows...  This might even make Wirt happy (a hard
>  thing to do...):  He and others might be able to more-
>  or-less do POSIX stuff in Linux land;  and "pure" MPE
>  and IMAGE stuff under MPE.   :-)

No, I'm actually very easy to please. I just don't like bad design. What I'm
arguing for takes away nothing; in contrast, it greatly adds capability while
simplifying MPE.

In essence, MPE's :SH command would simply be redirected to the underlying
Linux shell. Doing this solves a number of problems:

(i) the porting problem simply disappears. Porting becomes unnecessary. If
"whatever" exists in Linux, it now also exists in MPE. How do you tell bad
design? By the price you pay your consultants or the awards that you give
your volunteers. If the design was proper in the first place, neither the
consultants nor the volunteers would have ever been necessary. MPE subsumes
Linux in the emulator design that Gavin and I have been talking about since
the beginning, but the emulator is constructed so that the version of Linux
becomes irrelevant to the operation of MPE.

(ii) the legal problems associated with whatever proprietary code now exists
in the MPE POSIX shell would also disappear. It would no longer be reachable
and thus never expressed. Nor is any "un-intertwining" engineering effort
necessary. The POSIX code would remain in MPE, much as the code to build
teeth in birds is still present in their DNA, but it wouldn't be expressed,
just as the teeth code in birds hasn't been expressed in 50 million years.

Wirt Atmar