Funny Winboard bug

[hgm]

[Event "Computer Chess Game"]
[Site "FOM-RHKA8J2A5WY"]
[Date "2007.10.24"]
[Round "-"]
[White "umax4_8w"]
[Black "umax4_8w"]
[Result "1-0"]
[TimeControl "10"]

1. d4 d5 2. Nc3 Nc6 3. Bf4 Bf5 4. e3 Nf6 5. Bd3 Bxd3 6. cxd3 e6 7. Qb3 Na5
8. Qa4+ Nc6 9. Qb5 Qc8 10. O-O-O Be7 11. Nf3 O-O 12. h4 Nh5 13. Bh2 f5 14.
Ne2 a6 15. Qa4 Bf6 16. Nf4 Nxf4 17. Bxf4 Rb8 18. Qb3 Qd8 19. Qa3 Ra8 20.
Qb3 Rb8 21. Qa3 Be7 22. Qb3 Bd6 23. Bxd6 Qxd6 24. Rdf1 f4 25. a4 h5 26. Re1
Ne7 27. Ne5 Nf5 28. Qd1 fxe3 29. fxe3 c5 30. Qxh5 cxd4 31. exd4 Rbc8+ 32.
Kc1d1 Qb4 33. Qg4 Qb3+ 34. Kd1e2 Qxb2+ 35. Ke2f1 Ne3+ 36. Kf1g1 Qf2+ 37.
Kg1h2 Nxg4+ 38. Kh2h3 Nxe5 39. Rxe5 Qd2 40. a5 Rc2 41. Rg1 Rf1 42. Rxe6
Rxg1 43. Rg6 Rh1+ 44. Kh3g4 Qxg2+ 45. Kg4h5 Qf3+ 46. Rg4 Rxh4+ 47. Kh5xh4
Rh2+ 48. Kh4g5 Qf6+
{False win claim: 'Black mates'} 1-0

[d] 6k1/1p4p1/p4q2/P2p2K1/3P2R1/3P4/7r/8 w - - 3 49

Winboard does have a mate-test routine, mainly for determining if the SAN should have a + or # at the end. In Winboard_F I now use this mate test to adjudicate the game, (in case the engines don't claim), or to intercept false claims by engines.

However, the MateTest routine does not consider the above position checkmate: Winboard it annotates it with + in the PGN, and hence it rejects the claim by the engine, despite the fact that white is very checkmated.

Do you see why?

[Dirt]

It must think Kxf6 is a legal move because the pawn would be pinned by the rook.

[hristo]

[Event "Computer Chess Game"]
[Site "FOM-RHKA8J2A5WY"]
[Date "2007.10.24"]
[Round "-"]
[White "umax4_8w"]
[Black "umax4_8w"]
[Result "1-0"]
[TimeControl "10"]

1. d4 d5 2. Nc3 Nc6 3. Bf4 Bf5 4. e3 Nf6 5. Bd3 Bxd3 6. cxd3 e6 7. Qb3 Na5
8. Qa4+ Nc6 9. Qb5 Qc8 10. O-O-O Be7 11. Nf3 O-O 12. h4 Nh5 13. Bh2 f5 14.
Ne2 a6 15. Qa4 Bf6 16. Nf4 Nxf4 17. Bxf4 Rb8 18. Qb3 Qd8 19. Qa3 Ra8 20.
Qb3 Rb8 21. Qa3 Be7 22. Qb3 Bd6 23. Bxd6 Qxd6 24. Rdf1 f4 25. a4 h5 26. Re1
Ne7 27. Ne5 Nf5 28. Qd1 fxe3 29. fxe3 c5 30. Qxh5 cxd4 31. exd4 Rbc8+ 32.
Kc1d1 Qb4 33. Qg4 Qb3+ 34. Kd1e2 Qxb2+ 35. Ke2f1 Ne3+ 36. Kf1g1 Qf2+ 37.
Kg1h2 Nxg4+ 38. Kh2h3 Nxe5 39. Rxe5 Qd2 40. a5 Rc2 41. Rg1 Rf1 42. Rxe6
Rxg1 43. Rg6 Rh1+ 44. Kh3g4 Qxg2+ 45. Kg4h5 Qf3+ 46. Rg4 Rxh4+ 47. Kh5xh4
Rh2+ 48. Kh4g5 Qf6+
{False win claim: 'Black mates'} 1-0

[d] 6k1/1p4p1/p4q2/P2p2K1/3P2R1/3P4/7r/8 w - - 3 49

Winboard does have a mate-test routine, mainly for determining if the SAN should have a + or # at the end. In Winboard_F I now use this mate test to adjudicate the game, (in case the engines don't claim), or to intercept false claims by engines.

However, the MateTest routine does not consider the above position checkmate: Winboard it annotates it with + in the PGN, and hence it rejects the claim by the engine, despite the fact that white is very checkmated.

Do you see why?

Kxf6 and then a the pawn is 'pinned'? hahaha

[hgm]

Yes, you got it!

Difficult to fix, though; the way the mate test works in Winboard is very cumbersome.