Re: [gnugo-devel] test results after removing Pattern CD106

[kevin yong]

Hi, Evan:

  i ran 2 full regression tests before/after pattern
CD106 from patterns.db. it correct the problem from
13x13b:27. this time it gives right answer C4. the
removal of pattern CD106 has no effects on other test
results.

  all i did was quite simple: comment out pattern
CD106. run each .tst file, and output into 2 log
files:
log1 hold all detailed log from inside gnugo, while
log2 only holds the result of output from command
input, like:
input:

# B2 inefficient way to live.
loadsgf games/mertin13x13/GMGN2.sgf 92
27 gg_genmove white
#? [C4]*

output:

=27 C4

i only run 'diff' on the log2 files. among all log2
file for all .tst file, only difference found is 

107c107
< =27 C4
---
> =27 B2
211c211
< =10003 48662488
---
> =10003 48662491

and i run 'diff' on the patterns.db before/after
modification:

11456,11457c11456,11457
< #Pattern CD106
< ## db added (3.1.7)
---
> Pattern CD106
> # db added (3.1.7)
11459,11462c11459,11462
< #?x?          make an eye (or die trying!)
< #O*O
< #O.O
< #oOX
---
> ?x?          make an eye (or die trying!)
> O*O
> O.O
> oOX
11464c11464
< #:8,d
---
> :8,d
11466,11469c11466,11469
< #?x?
< #O*A
< #OaA
< #bOX
---
> ?x?
> O*A
> OaA
> bOX
11471c11471
< #; does_defend(*,A)
---
> ; does_defend(*,A)
11473,11474c11473,11474
< #> replace(a,*);
< #> if (!o_alive_somewhere(b)) replace(b,*)
---
> > replace(a,*);
> > if (!o_alive_somewhere(b)) replace(b,*)

  i m using gnugo3.3.17

  like to hear your advice.

kevin.




______________________________________________________________________ 
Post your free ad now! http://personals.yahoo.ca