Mark Klein writes:

> If you listen to the FSF and CYGNUS, you'll hear them touting that you
>  should write your software to a compiler environment such as GCC. Then,
>  if GCC and the requisite tools get ported to another environment, you
>  can then easily move software to that platform.
>
>  I'm beginning to believe in that, too.

I would wholeheartedly agree with Mark -- if I weren't getting old.

I remember well all of the great noise about ANSI-standard FORTRAN and COBOL
in the late 1960's. These languages were "mandated" to be exactly identical
on all platforms. Heck, I even remember the original reason for the
development of PASCAL in the early 1970's, an absolutely,
truly-for-the-first-time, designed-from-scratch platform-independent
upper-level language that would be the be-all and end-all of all future
portability problems. PASCAL would be absolutely the same on all platforms
because it would generate only P-code, an intermediate, virtual-processor
code that would then be object-code translated into the appropriate target
processor's native-mode code. The upper level shell of PASCAL would be
absolutely identical on all platforms.

The very first time that someone wrote a PASCAL compiler (MODCAL anyone?)
that skipped the P-code step and generated native target code, optimized for
the current processor of choice with processor-appropriate extensions,
compatibility went out the window.

It seems that even those who remember history are doomed to repeat it. PASCAL
is now remembered for merely being the first language to introduce structured
programming -- and for a bit of time, made the elimination of GOTOs into a
national religion. JAVA, the "write-once, run-anywhere" language, a phrase
almost identical to PASCAL's original messge, may soon similarly be
remembered for only have a better garbage collection mechanism than C++.

Wirt Atmar