Ticket #201 (new defect)

Opened 4 years ago

Last modified 3 years ago

Inconsistent boundary values

Reported by: gunnar Owned by: gnugo
Priority: normal Milestone: 3.9.x
Component: source Version:
Severity: normal Keywords:
Cc: patch: yes

Description

In the attached game at move 56,











owl_defend F2 incorrectly finds an owl defense at F3, whereas owl_does_defend F3 F8 correctly says 0. However, when generating a move, owl_defend F2 precedes the owl_does_defend F3 F8 call and the latter finds a cached success result from the former search.

Tracking this down turns out to be a difference in owl boundary values, specifically for the J4 stone.

Attachments

239160.sgf Download (1.4 KB) - added by gunnar 4 years ago.
Game record from CGOS.
gunnar_7_12.11.diff Download (2.6 KB) - added by gunnar 4 years ago.
Mark both goal and boundary in sgf output of owl/semeai reading.

Regression Results

Attachment Rev. PASS FAIL Nodes Status
gunnar_7_12.11.diff Download never tested

Change History

Changed 4 years ago by gunnar

Game record from CGOS.

Changed 4 years ago by gunnar

Mark both goal and boundary in sgf output of owl/semeai reading.

comment:1 Changed 4 years ago by gunnar

  • patch set

The attached patch gunnar_7_12.11 adds better debugging support for the owl boundary, such that owl boundary value 1 is marked by a circle in the sgf files from owl/semeai reading and owl boundary value 2 is marked by a triangle. The goal is as before marked by an sgf "Mark".

comment:2 Changed 3 years ago by gunnar

  • Milestone changed from 3.8 to 3.9.x
Note: See TracTickets for help on using tickets.