[Bug-gnubg] Money session bug ?

[Massimiliano Maini]

Hi all,

I tried to play a few money sessions
with gnubg and I have encountered
a couple of problems (first is minor,
second not IMO).

1 - When I double and gnubg beavers,
if I ask for a Hint on the take/pass
I get an empty dialog with an OK button.
I know gnubg doesn't analyze
beavers, but a simple message (as the
one displayed in the analysis panel)
would be better.

2 - I have a very short money sessions
for which the stats are completely
broken. Here's the ".mat"
file of the session (I have the sgf with the
analysis, but the mailing list may
filter it, mail me if you want it):

[play is intentionally silly here]

*************************************************************************

 0 point match

 Game 1
 gnubg : 0      
               MaX : 0
  1)        
                    46:
13/9 13/7 
  2) 45: 13/9 13/8    
           Doubles => 2
  3)  Doubles => 4  
             Takes
  4)  Takes    
                 

*************************************************************************


Now if I analyse the session (supremo),
I get the following :

         
                     
      gnubg              
 MaX                 
Chequer Play Statistics:
Total moves        
                    1
                   1
                
Unforced moves      
                   1
                   1
                
Unmarked moves      
                   0
                   0
                
Moves marked good      
                1    
               0    
            
Moves marked doubtful    
              0      
             0      
          
Moves marked bad      
                 0  
                 1  
              
Moves marked very bad    
              0      
             0      
          
Error rate (total)      
             +0.000      
        -0.086 ( -0.086)    
Error rate (per move)    
              +0.0      
         -86.4 ( -0.086)    
Chequerplay rating      
             Supernatural    
    Awful!              

Luck Statistics:
Rolls marked very lucky    
            0        
           0        
        
Rolls marked lucky      
               0    
               0    
            
Rolls unmarked      
                   1
                   1
                
Rolls marked unlucky    
               0    
               0    
            
Rolls marked very unlucky    
          0          
         0          
      
Luck rate (total)      
              -0.262 ( -0.262)  
  +0.045 ( +0.045)    
Luck rate (per move)    
             -262.5 ( -0.262)  
   +45.2 ( +0.045)    
Luck rating        
                  Haaa-haaa
           Good dice, man!    


Cube Statistics:
Total cube decisions    
               2    
               2    
            
Close or actual cube decisions  
       1            
       2            
    
Doubles        
                     
  1                  
 1                 
Takes          
                     
  0                  
 1                 
Passes        
                     
   0                
   0                

Missed doubles (below CP)    
          0          
         0          
      
Missed doubles (above CP)    
          0          
         0          
      
Wrong doubles (below DP)    
           1 (-0.237 ( -0.237))   1
(-0.237 ( -0.237))
Wrong doubles (above TG)    
           0        
           0        
        
Wrong takes        
                    0
                   0
                
Wrong passes        
                   0
                   0
                
Error rate (total)      
             -0.237 ( -0.237)  
  -0.237 ( -0.237)    
Error rate (per cube decision)  
     -236.6 ( -0.237)     -118.3 ( -0.118)  
 
Cube decision rating    
             Awful!      
        Awful!            
 

Overall Statistics:
Error rate (total)      
             -0.237 ( -0.237)  
  -0.323 ( -0.323)    
Error rate (per decision)    
        -118.3 ( -0.118)     -107.7 ( -0.108)
   
Equiv. Snowie error rate    
         -118.3          
    -161.5              
Overall rating      
                 Awful!  
            Awful!        
     

There's something wrong with the Cube
stats for player gnubg: gnubg
playing level was the same as the analysis
level (supremo), how can
he has a wrong double ? 
Notice that the supposed error is -0.237,
exactly the same as MaX error
(the error that triggered the beaver)
... like if player B error (on the
double) has been counted for player
A too (on his beaver) ...

This happens with both 0.15 and 0.16.

MaX.