-=> On 07 Oct 98  06:31:00 you wrote to me <=-

Hi Jasen,

JB> thanks, but what does this mean? (from from a run with penta.cfg)

JB>  3183 tests passed. Solution 5:            ignore the 3183 it's wrong.
JB> +-------------------+---------------+---+   the 5 is correct.

[Faulty solution snipped]

JB> Do I have a "dud" copy?

No, just a dud compiler.  :) I keep forgetting that some people are still
using 16 bit integers.  Sorry about that, everyone.

JB> The "tests" counter and/or printf needs to be changed to
use longints too,
JB> but I think I can do that one myself.

Yes, as you guessed, the problems you're having are due to my use of int in
a few places that require long int, or better.

The prime culprits are shapemap and newmap in fillboard().  These integer
are used as bitmaps of which shapes have been used so far.  Since penta.cfg
has 18 shapes, it overflows a 16 bit integer.

The other troublemakers are in blocks.c. Fortunately, they only cause
cosmetic problems, like the "3183 tests passed".

/* Count found solutions */
int solvecount;

/* Count succesfully tested pieces */
int testcount;

/* How often to report shape tests */
int maxtest;

If you change the program to make these variables into long integers, it
should all work properly.  If it _still_ gives weird results, I'll try
compiling it myself under 16 bit mode.

Thanks for the feedback,

Michael Stapleton of Graphic Bits.

 * AmyBW v2.15b7+ *
... This tagline is encrypted
--- AdeptXBBS v1.11z (FREEWare/2)