odd behavior

[Will Singleton]

I usually test at time 1 inc 1, but today I ran a tournament using pswbtb (winboard 441, windows 7) at 40 moves in 2 minutes. But it still used time 1 inc 1. So I tried a different computer. Same thing.

I don't know how to set up a debug file to see what's happening. I assume using winboard directly would work, so the issue must be with pswbtm.

Will

[Will Singleton]

I should mention that I use start positions from epd "silver.epd"

[H.G.Muller]

Are you sure you altered the TC in PSWBTM (in the tournment setup dialog), and not in WinBoard?

[F. Bluemers]

[Graham Banks]

Odd behaviour? I thought you were talking about CCC for a moment.
Hope you're having a nice break, Will. How is progress on Amateur these days?

Cheers,
Graham.

[Will Singleton]

Yes, of course. Here's some more info. Note that while the details windows shows 40/3, the pgn shows 180+1. I also notice the event tag in the pgn was left over from a previous tourney, I didn't specify one for this tourney. Seems like a clue.

>from the "details" window:

295x34 Gauntlet
20 games per pairing
Time Control: 40 moves in 3 minutes
Pondering is off
Starting Positions: C:\chess\programs3\winboard_441\PSWBTM\start positions\silver.epd
1 participants

Pairing 1
Game 1

Games Played: 0
Games Left: 20

PGN File: C:\chess\programs3\winboard_441\PSWBTM\games\null.pgn


>pgn from first game, you can see the tc is 180+1 (!):

[Event "286nopt"]
[Site "DAD-PC"]
[Date "2009.12.30"]
[Round "1.1"]
[White "yace"]
[Black "295x34"]
[Result "1/2-1/2"]
[TimeControl "180+1"]
[FEN "1rbq1rk1/1pp2pbp/p1np1np1/4p3/2PPP3/2N1BP2/PP1Q2PP/R1N1KB1R w KQ e6 0 1"]
[SetUp "1"]
[Number "1"]

{--------------
. r b q . r k .
. p p . . p b p
p . n p . n p .
. . . . p . . .
. . P P P . . .
. . N . B P . .
P P . Q . . P P
R . N . K B . R
white to play
--------------}
1. d5 Ne7 2. Be2 Nh5 3. O-O f5 4. Nd3 b5 5. cxb5 axb5 6. Nb4 Bd7 7. Rac1
Rf7 8. Kh1 fxe4 9. fxe4 Nf6 10. Rf3 Rb7 11. Rcf1 Qf8 12. Bg5 h6 13. Bxf6
Rxf6 14. Rxf6 Bxf6 15. Qd1 Kg7 16. a3 h5 17. g4 hxg4 18. Bxg4 Bxg4 19. Qxg4
Qc8 20. Qe2 Qe8 21. Qd2 Bh4 22. Qe2 Bg5 23. Qg2 Bf6 24. Qd2 Bh4 25. Qe2 Bg5
26. Qg2 Bf6 27. Qf3 Bg5 28. b3 Bf4 29. Nd3 g5 30. Qg4 Ng6 31. Nb4 Bd2 32.
Rf3 Nh4 33. Rh3 Qf7 34. Ne2 Qf2 35. Nd3 Qf1+ 36. Ng1 Kf6 37. Qe6+ Kg7 38.
Qd7+ Kh6 39. Rg3 Qf6 40. Qg4 Ra7 41. Qe2 Bf4 42. Nf2 Qf7 43. Rc3 Rxa3 44.
Ng4+ Kg7 45. Nf2 Qd7 46. Qb2 Ra7 47. Qe2 Kf8 48. Rc6 Ra1 49. Qh5 Ra7 50.
Nfh3 Qg7 51. Qg4 Ra8 52. Rc2 Ke8 53. Ra2 Rb8 54. Nxf4 exf4 55. Nh3 Qc3 56.
Qe6+ Kf8 57. Qh6+ Ke8 58. Qh5+ Kf8 59. Qh6+ Ke8 60. Qh5+ Kd7 61. Qg4+ Ke8
62. Qe6+ Kf8 63. Qh6+
{Draw by repetition} 1/2-1/2


>from tournament.pswbtm:

C:\chess\programs3\winboard_441\PSWBTM\games\null.pgn


295x34
C:\chess\programs3\amateur\amateur295x34.exe

C:\chess\programs3\winboard_441\PSWBTM\start positions\silver.epd
1
0
20
2
3
40
0
0
0
20
0
1
0
0
0

>from ntls.pswbtm:

C:\chess\programs3\winboard_441\PSWBTM\games\null.pgn


C:\chess\programs3\winboard_441\PSWBTM\start positions\silver.epd
0
0
0
20
2
3
40
0
1
0
1
1
0
1

[H.G.Muller]

OK, I think I understand what happens:

Probably in your winboard.ini there is the setting /timeIncrement=1. Now my gue it that PSWBTM, when yuo play 40/3, only pases the -mps 40 -tc 3 options to WinBoard, and no increment. So the increment that was read from the ini file remains in force. And when that is >= 0, WinBoard plays incremental TC, and ignores the mps setting. To play a classical TC, /timeIncrement has to be set to -1. Make sure this is what is in the winboard.ini file.

I guess this should be counted as a shortcming of PSWBTM; it scould always define a -timeIncrement, and use -1 in case of a classical TC.

In your initail poste you said it was playing 1+1 when you were asking for 40/2. This should never be possible, s it would require a different value for -tc. But in the example you now give, they are both 3 min, and then this explanation makes sense.

/timeIncrement=-1 in the winboard.ini should always work, as in the case of an incremental TC PSWBTM will overrule it by passing a command-line option to WB to overrule it with the selected increment. (And PSWBTM will always use -saveSettingsOnExit false, so it will always stay -1 in the ini fle even if you run an incremental TC.)

[Will Singleton]

You got it, thx. I should have known, as it's always been like that with winboard. And might I suggest that you consider fixing it in the next release of winboard. It's been discussed in the past. Why should the default value of inc survive the mps argument?

Thanks again.

Will

[Will Singleton]

Odd behaviour? I thought you were talking about CCC for a moment.
Hope you're having a nice break, Will. How is progress on Amateur these days?

Cheers,
Graham.

Hi Graham,

Progress is hard to come by. It's a fun hobby, though. Things that work for some programs, like dann's smooth null, make mine play worse. Seems like it's the interaction of heuristics that make the difference.

Will

[H.G.Muller]

You got it, thx. I should have known, as it's always been like that with winboard. And might I suggest that you consider fixing it in the next release of winboard. It's been discussed in the past. Why should the default value of inc survive the mps argument?

This is hard to fix, because reading the option values is dne by an automated proces. (In Board the options are even processed by the X-toolkit library.) So by the time any specific code starts to look at the options, and could recognize conflicting options and reslve the conflict, the information on the order in which the options appeared is no longer available.

This is really a PSWBTM bug. It should fuly specify the TC, giving all three of (mps, tc, inc) always. So I cannot fix it by releasing a new WinBoard.