[Bug-gnubg] Bug: Race theory, dice seed

[Christopher D. Yep]

Win XP
030923 (1.1395), both panel and non-panel version
(Problem also occurs in older versions of gnubg)

1. Set up any bearoff position such that both players have all their 
remaining checkers on the 1-6 points.
2. Select player 0 - gnubg, player 1 - human (on roll) [not necessary for 
bug, but makes it easier to verify]
3. Execute (either through the GUI or directly) a command that shows the 
race theory window, i.e. "show Kleinman", "show thorp", "show onechequer", 
or "show onesidedrollout".  Click "OK" to exit the race theory window.
4. Now roll for player 1 (human).  The roll will always be 5-4, regardless 
of the dice seed.
5. It's now gnubg's turn; it will roll 3-1.

The following dice rolls (if the game is continued without invoking the 
race theory window again) are 5-5, 5-2, 3-3, 6-5, 5-2, 6-4.

If steps 3-5 are repeated indefinitely (i.e. do in order steps 1, 2, 3, 4, 
5, 3, 4, 5, 3, 4, 5, ...), player 1 and player 0 will roll 5-4 and 3-1 
indefinitely (until the game ends).

I tested the above with older versions of gnubg.  The same behavior occurs 
(sometimes with a different dice sequence though).

With versions 030727 (1.1227), 030801 (1.1233), 030805 (1.1245), and 030813 
(1.1260) the dice sequence is 5-1, 6-3, 3-1, 4-2, 2-2, 4-2, 4-2, 2-1.

With versions 030818 (1.1282), 030824  (1.1292), 030904 (1.1345), and 
030923 (1.1395) the dice sequence is 5-4, 3-1, 5-5, 5-2, 3-3, 6-5, 5-2, 6-4 
(as reported above).

What is probably happening is that invoking the race theory window causes 
gnubg to set the dice seed to some given value.

I didn't do any testing on the no-gui version.

Chris