Hi,

as the Mersenne Twister algorithm for producing pseudo random naumbers is
known to be quite good, I thought it would be not such a bad idea to port
this from C to Xbase++ 

Because I am not very good at the C language and I know that there are some
around here that *are*, I attach the code for peer review (including the
original C code and documentation).

BTW, the original C code can be found on the official MT-website:

http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html

Ahhmmm... before I forget the code relies on the not-so-well-documented
functions BOR/BAND/BXOR for bitwise boolean operations - if anyone has
plain Xbase++ versions of those functions... 

Thomas Braun

Thomas Braun wrote:

> Because I am not very good at the C language and I know that there are some
> around here that *are*, I attach the code for peer review (including the
> original C code and documentation).

Sorry - the zip file could have been 50% smaller... I forgot to remove the
.exe and .obj files before zipping

Thomas

Thomas,

> Ahhmmm... before I forget the code relies on the not-so-well-documented
> functions BOR/BAND/BXOR for bitwise boolean operations - if anyone has
> plain Xbase++ versions of those functions... 

You mean like #78 on my site? 

Regards,

Phil Ide

*******************************************
*   Xbase++ FAQ, Libraries and Sources:   *
*   goto: http://www.idep.org.uk/xbase    *
* --------------------------------------- *
* www.xodc.org.uk - openSource Dev-Center *
*******************************************

So what if I am a modem junky?

Hi Thomas,

Here you are, the program functions MtRand() and SMtRand() were derived 
from the Mersenne Twister algorithm MT19937.c; the work of Makoto 
Matsumoto and Takuji Nishimura.

All Xbase++ code but not using Phil's #78   Probably much smarter 
coding than mine, so I'll pay #78 a visit... 

I like the MT algo, however, what I posted in the original "Random()" 
thread is much simpler, fast and effective.

Regards, Greg

Greg Doran wrote:

> I like the MT algo, however, what I posted in the original "Random()" 
> thread is much simpler, fast and effective.

Greg,

depending on what you need the rendom numbers for, simple may not be what
you need  But the MT algorithm is not suitable for real crypto
applications anyway...

Thomas