[ggo-discussion] race condition

Usually I can play gnugo via ggo fine, but under heavy load (e.g.
running a compilation in the background) something locks up: even though
it is gnugo's move it does not continue. Since gnugo does not consume
any CPU time I assume that it has computed its move but ggo does not
display the result.

If I play very slowly this does not happen, so it must be a race
condition somewhere.

More detailed observations: Under high load I see the following after
gnugo's last move: first I get the half-transparent stone marker of the
opposite color, only a moment later I get the marker in the correct
color. If I play while the marker is in the wrong color or very shortly
afterwards, ggo always locks up (i.e. I never get the response move,
undo/pass etc. do nothing, new game still works).

RedHat Linux 9 + all patches (kernel 2.4.20-19.9)
GNU Go Version 3.4
gGo 0.4

I had the same issue with GnuGo 3.2 and gGo 0.3.8.

Best,
Volker

Other related posts: