Other than ease of solution, a future-proof test suite

[Dann Corbit]

The duplicate is (of course) my fault.
In the actual game,
[pgn][Event "Springer-Berlin Club Ch"]
[Site "?"]
[Date "1922.06.02"]
[EventDate "1922.??.??"]
[Round "?"]
[Result "1-0"]
[White "Kurt Richter"]
[Black "Erwin Kipke"]
[ECO "B00"]
[WhiteElo "?"]
[BlackElo "?"]
[PlyCount "30"]

1. d4 d5 2. Nc3 Nc6 3. e4 dxe4 4. d5 Ne5 5. f4 exf3 6. Nxf3 Nxf3+ 7. Qxf3 Nf6
8. Bg5 Bd7 9. O-O-O g6 10. Bxf6 exf6 11. Re1+ Be7 12. Qxf6 Rg8 13. d6 cxd6 14.
Nd5 Be6 15. Rxe6 1-0[/pgn]

queenside castling is still possible for black. Hence, technically, the 'q' is correct, though it has no effect on choosing the best move.

I also collected the position from wtharvey.com, which, on this page: https://www.wtharvey.com/mcmxxii.html
has this:
Kurt Richter vs Kipke, Berlin, 1922
r2qk1r1/pp2bp1p/3pbQp1/3N4/8/8/PPP3PP/2K1RB1R w - - 0 1
[ Rxe6 if fxe6 Bb5+ mates ]

which lacks the castle option in the EPD string. I have a unique index on the EPD in my database, but if the position is given both with and without castling rights, it is hard to know which is correct most of the time. Bad e.p. markers are far worse, since the EPD standard demands that spurious EPD markers are present (even if not capture is possible or if the capturing pawn is pinned). I have spent some work cleaning those up since they are detectable, but it is hard to catch them all, especially with new data entering my database every day.

[peter]

Nevermind, I could have seen it on my own, just one after another with almost same string in uncommented suite, just looked at the commented epd always mainly, which I imported to .cbh too. Tried to copy- paste those strings here too, but that gave too many signs for one posting. Yet replaced it now because it's not of different (second one) challenge to the engines, regards

[Jouni]

Houdini 1.5a from 2011(!) is still impressive mate finder. It got 886/1000 so 100 more than Stockfish .