[gnugo-devel] CB20 in conn.db

[bump]

Gunnar pointed out that the pattern CB20 in conn.db looks
fishy. Since I can't remember what this pattern is
supposed to help I ran the regressions and found that
removing this pattern breaks strategy2:82 and trevor:460.

I remember trevor:460 but I find it hard to believe that
the intention of this pattern was to fix that
test. Actually I don't quite see how it works because it
does not affect the amalgamation of D15 and D13. With
or without the pattern, they are not amalgamated.

I agree with Gunnar that the pattern is bogus and it
should be taken down. It would be good if we could do
so without breaking strategy2:82 and trevor:460 because
they are both bad mistakes.

Dan