Re: [gnugo-devel] 261 games 3.1.31 vs 3.1.32

[Daniel Bump]

> > I've been playing 3.1.31 against 3.1.32 and have collected 116 games with
> > 3.1.32 as white, and 145 games with 3.1.32 as black.
> 
> If 3.1.31 is broken, couldn't 3.1.32 be just as broken?

Perhaps we should do a few test runs against 3.1.30 and 3.1.29. But
we have enough information to be confident that the current version
is a lot stronger than 3.0.

In the short run, the main question was whether to delay the release
because I had gotten worried by a few things I saw, the rating
slide, and bad runs on twogtp. (I didn't mention these concerns in
the list until yesterday.)  I've been reassured on both
counts. Gunnar thinks the rating slide is due to losses against
Lazarus and on time.

Also we need to play many games before a stable release in order
to find crashes as well as to test strength. Currently we can
crash in rare circumstances, since Arend gave an example in an
unrealistic circumstance where two replacement patterns are 
matched for a single move. So we need the patch at the end
of this message. Generally I don't think the current version
can crash very often.

> And in general, is playing against older versions of GNU Go a good test? 
> While testing Goban, I've been looking at a few games between different 
> versions of GNU Go, and they seemed strange to me (but I'm a weak 
> player). It would be nice to have other comparisons, like the game 
> series against Handtalk. I don't have easy access to a Windows PC 
> outside of work, but I'm thinking about investing in SoftPC for my Mac 
> just to match GNU Go against Smart Go or Many Faces or Go Professional.

About the strangeness of the games, current GNU Go will often start
to rescue a group, then abandon it because something else seems
bigger. Sometimes this works out OK because the opponent may invest
several moves to kill the stones in question. On the other hand
inconsistent play is a problem that needs to be fixed.

Dan

Index: engine/move_reasons.c
===================================================================
RCS file: /cvsroot/gnugo/gnugo/engine/move_reasons.c,v
retrieving revision 1.80
diff -u -r1.80 move_reasons.c
--- engine/move_reasons.c       18 Apr 2002 21:00:26 -0000      1.80
+++ engine/move_reasons.c       19 Apr 2002 14:18:11 -0000
@@ -1407,10 +1407,12 @@
   /* First check for an incompatible redistribution rule. */
   if (replacement_map[from] != NO_MOVE) {
     int dd = replacement_map[from];
-
-    /* Crash if the old rule isn't compatible with the new one. */
-    ASSERT1(dd == to || to == replacement_map[dd], from);
-    /* There already is a compatible redistribution in effect so we
+    /* Abort if the old rule isn't compatible with the new one.
+     * (But not in the stable release.)
+     */
+    if (0)
+      ASSERT1(dd == to || to == replacement_map[dd], from);
+    /* There already is a redistribution in effect so we
      * have nothing more to do.
      */
     return;