RE: [gnugo-devel] Bugfix

[Portela Fernand]

Dan wrote:

> > I prefer to know that we still have these problems.

> Could you explain the last statement?

It seem to me preferable to have fails in the regression tests so that
we are aware of the problems in the engine. If it suffices that the engine
considers a capture to make it misjudge a position, then yes, I prefer to
"know" that we have the problem (I currently don't know which one though,
I haven't looked into this stuff). The "still" refers to my assumption that
the failures disappeared unnoticed when a change happened in does_defend().
Else, I don't understand how that pattern constraint could have had this
comparison with the WIN constant.

/nando