Re: [gnugo-devel] trevor_1_15.6

[Trevor Morris]

At 10:47 PM 11/27/2001 +0100, you wrote:
>Trevor wrote:
>> http://www.public32.com/games/go/trevor_1_15.6
>>  - replaced abort() in mkeyes with fatal error messages.
>>  - lots of owl tuning.
>
>From the patch:
>
>> +Pattern A514
>> +# tm New Pattern (3.1.15)
>> +#   see also Pattern A615
>> +
>> +?XO.|        special corner position
>> +?.XX|
>> +?.*O|
>> +?...|
>> +
>> +:8,s,value(45)
>> +
>> +?XOb|        special corner position
>> +?CXX|
>> +?A*O|
>> +?...|
>> +
>> +; !oplay_attack(*,A,b,C,b)
>> +
>
>Is the comment wrong or is an edge missing?
comment wrong.

>
>> +Pattern A615
>> +# tm New Pattern (3.1.15)
>> +#   May be possible to generalize away from corner
>> +
>> +?XO.|        special corner position
>> +?.X.|
>> +?..*|
>> +?...|
>> +----+
>> +
>> +:8,s,value(45)
>> +
>> +?Xb.|        special corner eyespace collapse
>> +?.A.|
>> +?..*|
>> +?...|
>> +----+
>> +
>> +;!attack(b) || 1
>> +# FIXME: because stackp == 0, this attack check doesn't work right.
>
>I don't understand the comment. What is the problem?
Seemed to not work correctly.  I'll work up an example later.

>
>> +Pattern D629
>> +# tm New Pattern (3.1.15)
>> +
>> +|xX?        make an eye on the edge.
>> +|.OO
>> +|*.O
>> +|..O
>> +|..O
>> +|xX?
>> +
>> +:8,s,value(65)
>      ^
>
>Sacrifice pattern?
No, but no need to waste cycles testing the safety in this position.

>
>> +Pattern VA35
>> +
>> +?XX|
>> +X..|
>> +X*X|
>> +XO.|
>> +---+
>> +
>> +:8,s,value(45)
>> +
>> +
>> +?XX|
>> +XAB|
>> +X*X|
>> +XO.|
>> +---+
>> +
>> +; !oplay_defend(*,A,B,A)
>
>This is a strange constraint. As far as I can see it's equivalent with
>oplay_attack(*,A,A) or xlib(A) == 2.
yes, xlib(A)==2 is much cleaner.

On a different note, I've seen the lib helper fail, and think it may
be caching liberties, which it can't unless stackp == 0.

-trevor

>
>/Gunnar
>
>_______________________________________________
>gnugo-devel mailing list
>address@hidden
>http://mail.gnu.org/mailman/listinfo/gnugo-devel
>