Re: Symmetry corrections (was: RE: [gnugo-devel] code maintenance patch)

[Arend Bayer]

Gunnar wrote:
> Arend wrote:
> > This explanation isn't 100% accurate. The matching order doesn't change
> > (the DFA doesn't know about symmetries of patterns) in this case.
>
> Huh? Somehow the matcher has to come up with a match for the pattern
> at only one rotation instead of two. Why couldn't this affect the
> matching order?
The DFA does not know about symmetric patterns (it cannot as we use
the same DFA for all 8 spirals). Unnecessary symmetries are filtered out
afterwards in check_pattern_light, matchpat.c:1062 (I only recently
introduced that).

> > I am pointing this out because I think I should change that.
> Something like this (untested) patch?
I agree (although comparing floating point numbers there still makes
me nervous; I mean I would not even the crappiest processor expect to
turn 70.0000000 into 69.99999997 or s.th., but after what we've
experienced pre-3.2 I've lost all my faith there...)

Arend