[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [computer-go] Third KGS tournament: game-end protocol



I'm still trying to get an exact picture of what the server actually
does.  My first impression was that after the cleanup command and 2
passes, the game was completely over no matter what.  But I now
believe this process can continue indefinitely.

I'm also not sure about how the "undo" command works in the 
context of disagreement.   For instance ...

If I pass, then you pass, we are issued an undo command.  So it's as
if you didn't make the second pass.   Now you get sent genmove_cleanup
but what if you pass?   We again have 2 passes in a row.  Do I, the
first player to pass,  get an opportunity to kill your stones?   I have
to assume YES, but I'm not sure.   It would of course be unfair if I
didn't so I assume that it's ok.

I guess we can test all of this or look at out logs to look for specific
cases.  

- Don



   On Thu, 9 Jun 2005, Don Dailey wrote:

   >
   >>   Since everyone else is taking a stab at this, here is mine:
   >>
   >>   1. When B, W both pass, request final_status_list
   >>   2. If the results are same, we are done
   >>   3. If not, start requesting gen_move_cleanup (if it is not
   >>   implemented, assume a pass move)
   >>   4. After two more passes, all blocks are assumed alive.
   >
   > YES.  This is, in my opinion, exactly how it should be.

   i couldn't agree more. i always thought this is what was being proposed 
   all along, and couldn't understand what problems people see...
   how is this different from wms protocol by the way?

   and as to Don's previous mail - is undo sent to one of the programs on 
   disagreement after 2 passes? if so how is it that the undo command does 
   not have to be implemented, according to kgsgtp doc?

   oh, and as to 'assume everything is alive after 2 passes' - i would really 
   like to be able to have the same bot play against humans, and against 
   other bots... and humans are _not_ thrilled with a bot believing in kill all 
   before passing approach...

   > > - Don

   mike
   _______________________________________________
   computer-go mailing list
   computer-go@xxxxxxxxxxxxxxxxx
   http://www.computer-go.org/mailman/listinfo/computer-go/

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/