Re[2]: [gnugo-devel] connections.c

[Paul Pogonyshev]

>> i've discovered the following line in connection.c:
>> 
>> [67]   if (!EXPERIMENTAL_CONNECTIONS && (pattern->class & CLASS_C)) {
>> 
>> shouldn't macro EXPERIMENTAL_CONNECTIONS be changed to the
>> corresponding variable experimental_connections?

Gunnar wrote:
> Yes, it should. My mistake, I suspect.

since no patch appeared yet, here it is.

i believe someone said on the list that experimental connections was
tested on nngs and showed to be worse than the standard
implementation. couldn't it be connected with this line if it was
turned on by --experimental-connections option? (it was probably not,
but maybe it's better to check)

Index: connections.c
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/connections.c,v
retrieving revision 1.25
diff -u -r1.25 connections.c
--- connections.c       22 Jul 2002 16:49:26 -0000      1.25
+++ connections.c       31 Aug 2002 12:35:37 -0000
@@ -64,7 +64,7 @@
    * previously during find_cuts.
    */
 
-  if (!EXPERIMENTAL_CONNECTIONS && (pattern->class & CLASS_C)) {
+  if (!experimental_connections && (pattern->class & CLASS_C)) {
     for (k = 0; k < pattern->patlen; ++k) { /* match each point */
       /* transform pattern real coordinate */
       int pos = AFFINE_TRANSFORM(pattern->patn[k].x, pattern->patn[k].y,