Stockfish-1.5 Detects mate then changes mind?

[Jouni]

I haven't read whole thread , but FYI one position from Chest mate base;mate in 4:
[d]1B2n3/8/2R5/5p2/3kp1n1/4p3/B3K3/8 w - - bm #4;

Analysis by Stockfish 1.5.1 JA:

1.Ba7+
+- (4.56) Depth: 1 00:00:00
1.Bf4 Nef6 2.Rd6+ Kc3
+- (4.60) Depth: 2 00:00:00
1.Ba7+ Ke5 2.Re6+ Kf4 3.Rxe8 Kg5 4.Bxe3+ Nxe3 5.Kxe3
+- (8.56) Depth: 2 00:00:00
1.Ba7+ Ke5 2.Re6+ Kf4 3.Rxe8 Kg3 4.Bc5
+- (8.64) Depth: 3 00:00:00
1.Ba7+ Ke5 2.Re6+ Kf4 3.Rxe8 Kg3 4.Bd5 Nf6
+- (8.68) Depth: 4 00:00:00
1.Rc4+ Kd5 2.Rc8+ Kd4 3.Be6 Nc7 4.Bxc7 Nh6 5.Rd8+ Kc5 6.Kxe3
+- (10.26) Depth: 4 00:00:00
1.Be6 Nc7 2.Bxc7 Nf6 3.Rc4#
+- (#3) Depth: 4 00:00:00
1.Be6 Nc7 2.Bxc7 Nf6 3.Rc4#
+- (#3) Depth: 5 00:00:00
1.Be6 Nc7 2.Bxc7 Nf6 3.Rc4#
+- (#3) Depth: 6 00:00:00 6kN
1.Be6 Nc7 2.Bxc7 Nf6 3.Rc4#
+- (#3) Depth: 7 00:00:00 6kN
1.Be6 Nc7 2.Bxc7 Nf6 3.Rc4#
+- (#3) Depth: 8 00:00:00 7kN
.
.
.
1.Be6 Nc7 2.Bxc7 Nf6 3.Rc4#
+- (#3) Depth: 100 00:00:01 134kN

Jouni

[mcostalba]

.
1.Be6 Nc7 2.Bxc7 Nf6 3.Rc4#
+- (#3) Depth: 100 00:00:01 134kN

Jouni

Hi Jouni,

we think we have understood what the problem is, we are looking for a solution that does not adversely impact performance.

Thanks for your report
Marco

[Vinvin]

Good news, does that means will get a Sotckfish 1.5.2 soon ?

.
1.Be6 Nc7 2.Bxc7 Nf6 3.Rc4#
+- (#3) Depth: 100 00:00:01 134kN

Jouni

Hi Jouni,

we think we have understood what the problem is, we are looking for a solution that does not adversely impact performance.

Thanks for your report
Marco

[mcostalba]

Good news, does that means will get a Sotckfish 1.5.2 soon ?

I don't think so. This behaviour was present already in Stockfish 1.4 and none noticed, now it is under the spotlight and so a lot of small bugs are reported.

This is very good for us, but it is also important try to keep the releases to a bare minimum to avoid hassles to the testers, 1.5.1 was necessary due to Chess960, but I think this bug has already waited some months so can wait some months more until the next release