Re: [gnugo-devel] testing the 1-D boardcode: BOARD_OFFSET

[Arend Bayer]

On Sun, 28 Apr 2002, Daniel Bump wrote:

>
> Teun wrote:
>
> > If all goes well, the 1-D board should be handled through
> > the macros in liberty.h only. The idea behind this patch
> > is that handling of the 1-D board can be tested by
> > introducing a BOARD_OFFSET in liberty.h. BOARD_OFFSET=0
> > gives the standard situation. BOARD_OFFSET>0 can be
> > used for testing.
> >
> > Currently Gnu Go crashes in the first reading test when
> > BOARD_OFFSET = 1.

> It seems dangerous to define PASS_MOVE to be other than
> zero. It could be hard to track down all the places where
> this is assumed.
>
> If don't try to define PASS_MOVE to be nonzero, the patch
> still produces crashes however, the first in reading test 5.

I'd guess that there are numerous places where PASS_MOVE == POS(-1, -1)
is assumed.

Arend