[gnugo-devel] 3.1.33 VC inconsistencies

[Trevor Morris]

At 05:24 PM 4/19/2002 -0700, address@hidden wrote:

>> As of about the time you wrote that, here are the VC unexpected
>> results (updated on http://www.public32.com/regress ):
>
>I think you had results identical to the regressions for 3.1.32, is
>that correct?
>
>One thing you could try is to back out the patches in 3.1.32 one or
>two at at a time and try to figure out which one broke the platform
>independence.

The 3.1.32 regressions were not identical, per the thread:
  GNU Go 3.1.32 - VC inconsistencies 

However, the current differences were not present in 3.1.32.  I've not run
the full regressions since 3.1.32.

BTW, the cygwin results have no unexpected regression results.

I've noticed one thing so far that's odd.  Namely, aa_attackpats.c is different
btw. VC & cygwin builds.  You can see the VC version at:
  
http://www.public32.com/pcolon/builds/gnugo/2002-4-19/gnugo/patterns/aa_attackpat.c

The only other files that differ are eyes.c and the joseki files, but I believe
this is CR/LF differences only.  The differences in aa_attackpat.c look
substantive.

Secondly, here's the command I've run:
$ gq -o 2 -l games/incident258.sgf -L 147 --decide-dragon C4 -a -w -t -d0xa -o 
varsvc.sgf 2> xvc

(output available at http://www.public32.com/games/go/varsvc.sgf
and [ http://www.public32.com/games/go/xvc 
or  http://www.public32.com/games/go/xvc.gz ] )

This deviates immediately.  Breaking at the first trace line in owl.c line 
3004, here's the
list contents:

+       list->pattern_list[0].pattern->name     0x006f8cd0 "A1116"
+       list->pattern_list[1].pattern->name     0x006f8c60 "A1104"
+       list->pattern_list[2].pattern->name     0x006f8c90 "A1108"
+       list->pattern_list[3].pattern->name     0x006f8c90 "A1108"
+       list->pattern_list[4].pattern->name     0x006f8d48 "A1201"
+       list->pattern_list[5].pattern->name     0x006f8d18 "A1124a"
+       list->pattern_list[6].pattern->name     0x006f8d10 "A1124"
+       list->pattern_list[7].pattern->name     0x006f8d10 "A1124"
+       list->pattern_list[8].pattern->name     0x006f8940 "A505"
+       list->pattern_list[9].pattern->name     0x006f87a0 "A217b"
+       list->pattern_list[10].pattern->name    0x006f8798 "A217"
+       list->pattern_list[11].pattern->name    0x006f87a8 "A217c"
+       list->pattern_list[12].pattern->name    0x006f8690 "A101"
+       list->pattern_list[13].pattern->name    0x006f8d18 "A1124a"
+       list->pattern_list[14].pattern->name    0x006f8980 "A513"
+       list->pattern_list[15].pattern->name    0x006f8980 "A513"

VC picks A1124a.  cygwin picks A101.

I won't have any more time this weekend to look into it, but if a patch 
appears, I
may get a chance to try it out.

HTH,
Trevor