[Bug-gnubg] Time controls in Gnubg

[Stein Kulseth]

Hi everybody,

I have checked in a first implementation of gnubg with clocks on a 
separate branch
branched off the pre-rel-0-13-0 tag.
The branch is called pre-rel-0-13-0-tc and it configures with
  --with-timecontrol
as default which turns on the time control feature

How to use the time controls are described in the programs help functions, 
there's
even a small tutorial:
  show tctutorial
In this there are also some examples that you can add to your .gnubgrc 
file to have a few types of
time controls ready set up.

I have not tried it out on anything other than my own machine, running
Linux, and GTK 2. It works for me both with the GTK GUI and with the tty 
(-t)
interface, but the amount of testing that has gone into it is not 
overwhelming

Known missing features/Bugs
- Saving files with time info is not implemented (see below), and for now 
it actually does crash.
- Pausing the clock is not implemented.
- No GUI for setting clock parameters



About saving files with time info.

As far as I can see, the current sgf game format will have problems 
recording all info needed to record
a timed backgammon match.

The timing properties available are
TM - Time limit for game
OT - Timing description (free text)
WL, BL - Time left after the move.
OB, OW - Moves left (of byo-yomi)

We want to be able to:
- Start off the game with different time on each player's clock,
  The TM property applies to both player's, and the WL and BL properties
  are not gameinfo type. Suggestion:
  - Extend the MI property to include time left for each player,
      This property is extensible by design, also it is in th BG domain,
      so it doesn't impact the rest of the sgf format (however, other games
      might possibly also want different time limits in the gameinfo 
section)

- More accurately record the type of timing.
  The actual time left after each move can be coded with the existing 
properties
  but it would be more than nice to also record whether the timing type 
actually used
  was Bronstein, Fischer, plain etc. ...
  Again I think we should extend the MI.

- Enact point loss (actually opponent gain) during the game
  The result property RE only allows for one player to win points from the 
game,
  also it is implicitly indicated that this result is incurred at the end 
of the game.
  Suggestion:
  - introduce a setup property that changes score and records timing 
changes.
    (If this is a genaral setup property, this can also be used to impose 
penalties from TD
    in a tournament setting)


Feedback wanted on these issues.
Also on what process requirements there are on changing/extending the sgf 
spec.

--
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/