[gnugo-devel] current regression failures

[Gunnar Farneback]

I've checked the current regression failures with patches up to and
including gunnar_1_16.5.

Summary:
Only a few of the failures can be attributed to weaknesses in the
patches added post 3.1.15, and all of those have to do with the
atari_atari code. There's in particular no indication that there would
be any problems with the semeai patches (in non-experimental mode),
the optics patch, Arend's moyo patch, or the non-atari_atari parts of
Inge's patch. When it comes to atari_atari patches Inge's changes have
effectively been disabled by my revisions so the remaining problems
there are all on my account.


In the detailed listing below, the module within parenthesis is the
one which was involved in the change in status. I'd appreciate if
stronger players could review my questions and comments about the test
cases themselves.

atari_atari:1 (atari_atari)
Correct answers at L3 and N2 are missing.

neurogo:16 (atari_atari?)
I don't understand this test case. Can someone explain what it's
about?

strategy:27 (optics?)
Compared to 3.1.15, an owl attack on D2 is found. This together with
the fact that the proposed owl defense of D2 is bad (F2 rather than
C3) causes the failure.

strategy2:73 (semeai, moyo?)
To me it seems like E7 is very similar to F7. Any reason not to add it
to the correct answers?

strategy2:100 (atari_atari)
This is a failure for the atari_atari code but there's also something
fishy going on with the owl reading of O13 since it fails to find a
defense. This is nothing new, by the way, and will be added as
owl:254.

nicklas1:1510 (atari_atari)
This is a bad failure for the atari_atari code. This can be solved by
letting the atari_atari code return multiple defense points.

nicklas5:804 (moyo)
This is a case of several bad strategic values. That it passed before
and fails now is mostly coincidental.

trevor:381 (atari_atari)
This actually passed before due to a bug in the atari_atari code. Now
the main problem is with an incorrect reverse followup value.

strategy3:121 (atari_atari)
The comment to nicklas1:1510 applies here too.

global:26 (moyo)
Very shaky valuation. Mostly luck it passed before.

trevora:120 (atari_atari)
Various valuation mistakes. However, the previous move at E17 isn't
satisfactory and shouldn't be listed as correct. Or am I missing
something?

strategy4:166 (atari_atari)
Mistake by atari_atari. Returning multiple defense moves best
solution.

strategy4:179 (moyo)
Only luck it passed before.


Below follows comments on some failures which were new in 3.1.15:

global:41, vie:41
A9 wastes a ko threat but mostly looks okay to me. It suffices to
live.

nicklas1:1106
What's so bad about F6? 1 point reverse sente is maybe not the biggest
move on the board but it's also far from the worst. This test should
either be removed or changed to require the largest move on the board
(I guess this would be Q6 or some move around T12).

owl:136
I don't understand this test. Supposedly F14 would kill but black
responds at B12 and has two certain eyes. Am I missing something?

/Gunnar