Re: Jongware -> Building Galaxy Generator in C#

"Marcin Grzegorczyk" <mgrzegor@[EMAIL PROTECTED]
> wrote in message
news:fjpbkj$dtt$1@[EMAIL PROTECTED]
> MICHAEL.MCALEA@[EMAIL PROTECTED]
 wrote:
> > [...]
> > Thanks Jong, it works perfectly under borland C++ 5.5 which means its
> > a compiler problem *Grrrr*.
>
> I'd suspect non****table code rather than a compiler problem.  It's not
> that easy to write ****table C (especially when ****ting from assembly),
> so many details of C are implementation-defined or unspecified (or even
> undefined).

While a perfectly valid statement, it doesn't rule out programmer errors;
I
found one in Mike's code, and when fixed, it worked perfectly.

A /possible/ implementation flaw is when some compilers zero out new
variables
and others don't -- a very popular one in MSVC ("debug version runs but
release
code does not", as often seen on MS forums).
Your comment on the behaviour of rotl/rotr has been duly noted
(implementation
of rotl/rotr). With the singular exception of the rotl/rotr code all
worked
perfectly first go on a competitive C compiler (MSVC vs Borlands).
But other unexpected errors can be expected when people try my code on
other-endian or 64 bit machines -- or, gods forbid, on 8 bit machines.

I can only comment: "At least now it's in C instead of pure 386 assembly."
That
should at least open the door to other platforms and architectures.

[Jongware]