Ticket #9 (closed defect: fixed)

Opened 7 years ago

Last modified 6 years ago

Scoring crash

Reported by: gunnar Owned by: gunnar
Priority: normal Milestone: 3.7.8
Component: source Version:
Severity: blocker Keywords:
Cc: patch: yes

Description

 http://lists.gnu.org/archive/html/gnugo-devel/2005-04/msg00010.html

I haven't yet had time to investigate this, but I found the following
crash (run final_score in gtp mode on the sgf).  It is reproducible in
current CVS.

Evan


***assertion failure:
board.c:2645 - board[pos] == EMPTY near E3***

 (variation 59741)
   A B C D E F G H J
 9 . . . X X O . O . 9
 8 . . X X O O O O O 8
 7 . X O X X X O . . 7
 6 . . . X X O O O O 6
 5 . . . X + X X O X 5
 4 . . X . O X . X X 4
 3 . . X . X O X X X 3
 2 . . X X O O . . . 2     WHITE (O) has captured 0 stones
 1 . . . . X O . . . 1     BLACK (X) has captured 6 stones
   A B C D E F G H J
(;GM[1]FF[4]SZ[9]KM[0.0]GN[GNU Go 3.7.2 stepped on a bug]
;B[cg];W[gd];B[dd];W[fg];B[fc];W[gc];B[ed];W[cc];B[db];W[dg];B[cf];W[fb]
;B[cb];W[fd];B[dh];W[eh];B[ch];W[df];B[de];W[ef];B[fe];W[ee];B[ec];W[eb]
;B[ea];W[gb];B[ei];W[fi];B[ff];W[gf];B[eg];W[fh];B[ge];W[he];B[hf];W[if]
;B[hg];W[ef];B[gg];W[id];B[ig];W[fa];B[da];W[ib];B[ie];W[hd];B[if];W[]
;B[];W[];B[];W[ha];B[bc];W[hb];B[dc]
)
gnugo 3.7.2 (seed 0): You stepped on a bug.

Attachments

gunnar_7_8.4 Download (1.1 KB) - added by gunnar 6 years ago.
bugfixes in aftermath.c

Regression Results

Attachment Rev. PASS FAIL Nodes Status

Change History

comment:1 Changed 6 years ago by gunnar

  • Owner changed from gnugo to gunnar
  • Status changed from new to assigned

comment:2 Changed 6 years ago by gunnar

  • Milestone changed from 3.8 to 3.7.8

There are three issues conspiring to cause this crash.

  1. The D9 dragon is evaluated to status UNKNOWN.
  2. Line 673 of aftermath.c tests status != ALIVE, thereby counting D9 as something that should be taken off the board,
  3. The test at line 757 that a backfilling move to avoid self-atari really isn't a self-atari should first check that the point isn't occupied.

Changed 6 years ago by gunnar

bugfixes in aftermath.c

comment:3 Changed 6 years ago by gunnar

  • patch set

For issue 1, see ticket #35.

Issues 2 and 3 are fixed by the attached patch gunnar_7_8.4.

- bugfixes in do_aftermath_genmove()

comment:4 Changed 6 years ago by gunnar

  • Status changed from assigned to closed
  • Resolution set to fixed
Note: See TracTickets for help on using tickets.