Re: [Bug-gnubg] Extendable rollouts

[Jim Segrave]

On Sun 13 Jul 2003 (13:27 +0200), Øystein Johansen wrote:
> Nice work!
> 
> First bug report: If I start a cubeless rollout, and restart it cubeful, 
> the program crashes.

Can you give me a more detailed method of doing this? Changing
cubeful/cubeless should have no effect on restarts, as it should be
using the saved rolloutcontext, which won't have been changed.

I just tried stopping a cubeful rollout, changing the rollout setting
to cubeless and then resuming the rollout - nothing odd happened, 

I also tried rolling out a move cubeless, stopping, changing the
rollout settings to cubeful and resuming the rollout, again nothing
happened. 

Is it possible to give a position or move and the rollout settings
involved?


> The rollout progression in the main window status bar is now meaningless.

True - that's a bit of laziness on my part. When MoveListRollout
starts up a rollout, we need to get gnubg to complete redrawing the
board etc. before passing control to the rollout window. I had removed
the ProgressStartValue() call in MoveListRollout() and had the
following unplesant effect:

If the annotation window overlapped the board, then when you select
several moves to rollout and click rollout, the board and rollout
window move to the top and the rollout begins. But the area of the
board which was under the annotation window remains a grey blank until
the first rollout trial completes, which can be a while when rolling
out several moves. There's probably a better way to do this.

The main window status used to show which move was being rolled out
and progressed from 0 to n - 1 (or maybe 1 to n) when rolling out a
list of moves. Now it does trial 1 move 1, trial 1 move 2,
etc. Calling ProgressValueAdd would lead to some pretty absurd things
after the first trial. I'm open to suggestions as to what would be
more appropriate.


-- 
Jim Segrave           address@hidden