[Bug-gnubg] Re: Random problem with CLI version in 0ply analysis

[Massimiliano Maini]

Additional info: doing a 1ply analysis instead of 0ply,
on top of the random sgf:output differences I also get
random "crashes" on the assert on line 960 in eval.c
(CalculateHalfInputs):

          g_assert( n );

These crashes happen randomly, but only with >1
thread (just like the sgf/output random problem).
They happen with g threads and win threads, with
-O3 and -O0. In nogui exes only: gui exes looks OK.

MaX.