Subscribe
Notify of
guest
16 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Kenneth Solja
Kenneth Solja
6 years ago

Can you give some examples? I haven’t noticed any bugs ..

seetharaman
seetharaman
6 years ago
Reply to  Kenneth Solja

Ex.1.
White : Kf8 Qg1 Black: Kh8, Be5 #1. Popeye gives no solution ignoring 1. Qg8#. With Be5 moved to e4. Popeye gives single solution 1. Qg8#, but ignores 1.Qg7# !
Ex.2
white Pb2 Kc8 black Ka8 Ra1b1 #1 Popeye correctly gives 1.b2-a2 # If you replace Ra1 by black queen, Popeye gives ‘no solution’ even though the black queen cannot even now defend the mate (the queen cannot move as it has only pawn move after the available mate 1.b2-a2#).

dupont
dupont
6 years ago

In example 1, Qg8+ is not checkmate because of the parry 1…Bg7+ (and now the only legal move is 2.Ke8).

seetharaman
seetharaman
6 years ago
Reply to  dupont

You are right. I am sorry! 1.Qg8+ is not mate with black bishop on e5. I missed the point of the defence 1…Bg7. But with black bishop on e4 why is 1.Qg7# not mate? Am I missing anything?

dupont
dupont
6 years ago

You’re right, 1.Qg7# is checkmate with bBe4. This clearly illustrates that Popeye is bugy concerning BtB. WinChloe seems safe, although it doesn’t respect the rule concerning pawns on their first row – one should add the condition “Un Pion est impuissant sur sa première rangée”.

seetharaman
seetharaman
6 years ago
Reply to  dupont

Example 3: This is very interesting !!
white Pb2 Ke5 black Qh1 Ka8 Ra6a7 H#3.
One of the solutions given by Popeye is:
1.Qh1-b1 b2-b8=Q + 2.Ra7-b7 Qb8-c8 + 3.Rb7-b8 Qc8*a6 #
Obviously 2.Rb7 does not defend the check and is illegal !

Dominique Forlot
Dominique Forlot
6 years ago

Besides, for the same “example 3”: Popeye ignores some solutions as:

1.Qh1-b1 b2-c3 2.Qb1-c1 c3-c7 3.2.Qc1-h1 c7-c8=Q #

with a nice switchback of the Queen!

Dominique Forlot
Dominique Forlot
6 years ago

sorry for “bad typing”! Read :

1.Qh1-b1 b2-c2 2.Qb1-c1 c2-c7 3.Qc1-h1 c7-c8=Q #

Dominique Forlot
Dominique Forlot
6 years ago

another strange effect !
in the example 4 ( = example 3 with two knights instead of two Rooks )

white Pb2 Ke5
black Qh1 Sa7 Sa6 Ka8

Popeye find this solution ( with a dual )

1.Qh1-e4 + Ke5-a5 2.Qe4-b1 Ka5*a6 3.Sa7-b5 b2-h8=Q #

but ignore the rook promotion, also correct:

1.Qh1-e4 + Ke5-a5 2.Qe4-b1 Ka5*a6 3.Sa7-b5 b2-h8=R #
Popeye seems to consider 4.Qb1-h7?! like a defense ( as if it was just a bishop, not a queen )

Joost de Heer
Joost de Heer
6 years ago

Popeye 4.68 available at http://sanguis.xs4all.nl/popeye (only 32-bit version), which has fixed these errors, as far as I can see.

dupont
dupont
6 years ago
Reply to  Joost de Heer

It seems there are still errors with proof games (or a=>b problems).

Kenneth Solja
Kenneth Solja
6 years ago

It seems that all errors in Popeye are connected with Queen, at least I have got that impression ..

Thomas Maeder
Thomas Maeder
6 years ago

> It seems that all errors in Popeye are connected with Queen, at least I have got that impression ..

This is my impression as well. The reason is (as almost always in these cases) an optimisation: in orthodox chess and many fairy conditions it isn’t necessary to test for observations by queens in a separate step; Popeye rather follows bishop and rook lines and tests if there is a bishop or queen resp. rook or queen lurking.

This optimisation has to be turned off in FaceToFace and related conditions.

Popeye 4.69 does this; it correctly solves all the examples given above.

Cf.
https://sourceforge.net/projects/popeye-chess/files/Popeye/4.69/
http://www.matplus.net/start.php?px=1408299550&app=forum&act=posts&fid=it&tid=1508&pid=12540#n12540

seetharaman
seetharaman
6 years ago
Reply to  Thomas Maeder

Thanks Thomas for the quick response and correction!

Dominique Forlot
Dominique Forlot
6 years ago

Well done, for this reactivity and all the work upstream!