[computer-go] KGS tournament problem

[House, Jason J.]

With respect to the question below (what did housebot say was dead in
round 1?), housebot responded (correctly) with "B13 C13".  I had
checked this during the tournament.  When repeated problems (with
cleanup) occurred, I posted the (correct) final_status_list dead
responses from housebot/hbotsvn to the game discussion.  

It came as a surprise to me that the cleanup stuff was in cgoban3.  It
would be nice if there could be a more graceful handling of when one
bot doesn't support the cleanup command.  In the incorrect cases I saw
online, all the incorrectly alive stones were in atari.  It'd be nice
if they could be assumed dead.

-----Original Message-----
About HouseBot - WeakBot50k game in round 1:
http://files.gokgs.com/games/2006/10/8/HouseBot-WeakBot50k.sgf

Before cleanup:
FINEST: Got successful response to command "final_status_list dead": =
B13
J5
M5
Oct 8, 2006 12:12:03 PM org.igoweb.igoweb.client.gtp.P a
FINEST: Submitting stone status to server: Location = j5, dead = true
Oct 8, 2006 12:12:03 PM org.igoweb.igoweb.client.gtp.P a
FINEST: Submitting stone status to server: Location = m5, dead = true
Oct 8, 2006 12:12:03 PM org.igoweb.igoweb.client.gtp.P a
FINEST: Submitting stone status to server: Location = b13, dead = true
Oct 8, 2006 12:12:03 PM org.igoweb.kgs.client.gtp.GtpClient d
FINE: Dead stones all submitted to server
Oct 8, 2006 12:12:14 PM org.igoweb.igoweb.client.gtp.P a
FINEST: Submitting stone status to server: Location = j5, dead = true
Oct 8, 2006 12:12:14 PM org.igoweb.igoweb.client.gtp.P a
FINEST: Submitting stone status to server: Location = m5, dead = true
Oct 8, 2006 12:12:14 PM org.igoweb.igoweb.client.gtp.P j
INFO: Disagreement over tournament scoring. Switching to cleanup mode.


After cleanup where WeakBot50k did remove dead stones in its area:

FINEST: Got successful response to command "final_status_list dead": =
B13
Oct 8, 2006 12:12:22 PM org.igoweb.igoweb.client.gtp.P a
INFO: Cleanup mode has ended by passes. It will be assumed that all
dead
stones have already been removed.



There is 'feature' in kgsKgtp: you can mark opponent dead stones alive
or your live stones dead and opponent bot will accept that ;-)

Either above feature was used or as above message says, all stones are
assumed alive.

About initial disagreement: What does HouseBot say in log file as
response to final_status_list command?