Re: Poliglot and checkmate

[Matthias Gemuh]

Geschrieben von:/Posted by: Matthias Gemuh at 31 March 2004 18:52:32:
Als Antwort auf:/In reply to: Re: Poliglot and checkmate geschrieben von:/posted by: Fabien Letouzey at 31 March 2004 18:18:27:

Sometimes Poliglot wrongly(?) criticises checkmate:



I hope this report helps.

There is a bug in the engine ("mate in 6" is not a UCI-compatible mate announcement). Which one was it?
It is still helpful, when I try to accomodate buggy output from engines.
Fabien.

>>< ENGINE info depth 5 score mate in 6 moves time 5 nodes 1011 pv g7f8 h2g1
>>file "xboard.cpp", line 855, assertion "mate!=0" failed
>>


Hi Fabien,
I am impressed how cool you react concerning buggy UCI support in so many engines. Most adapter programmers would boil and curse .
Thanx for PolyGlot.
/Matthias.


BigLion + Taktix

[Fabien Letouzey]

Geschrieben von:/Posted by: Fabien Letouzey at 31 March 2004 18:59:58:
Als Antwort auf:/In reply to: Re: Poliglot and checkmate geschrieben von:/posted by: Matthias Gemuh at 31 March 2004 18:52:32:

Sometimes Poliglot wrongly(?) criticises checkmate:



I hope this report helps.

There is a bug in the engine ("mate in 6" is not a UCI-compatible mate announcement). Which one was it?
It is still helpful, when I try to accomodate buggy output from engines.
Fabien.

Hi Fabien,
I am impressed how cool you react concerning buggy UCI support in so many engines. Most adapter programmers would boil and curse .
Thanx for PolyGlot.
/Matthias.

>>>< ENGINE info depth 5 score mate in 6 moves time 5 nodes 1011 pv g7f8 h2g1
>>>file "xboard.cpp", line 855, assertion "mate!=0" failed
>>>
I expected many such problems.
On Linux I could test with only 4 UCI engines
PolyGlot is in DEBUG mode meaning assertions triggers all the time. I think this is irritating for testers, but there are good sides:
1) bugs are located in engines, they can be fixed for the future
2) I see what kind of output to expect, so that my parser tries to still get some info in the future
3) each problem is a potential one in PolyGlot so it's better to have a close look at them
I could make PolyGlot available in release mode, but that would only hide the problems.
Fabien.

[Fabien Letouzey]

Geschrieben von:/Posted by: Fabien Letouzey at 31 March 2004 19:03:04:
Als Antwort auf:/In reply to: Re: Poliglot and checkmate geschrieben von:/posted by: Fabien Letouzey at 31 March 2004 18:59:58:

Sometimes Poliglot wrongly(?) criticises checkmate:



I hope this report helps.

There is a bug in the engine ("mate in 6" is not a UCI-compatible mate announcement). Which one was it?

>>>>< ENGINE info depth 5 score mate in 6 moves time 5 nodes 1011 pv g7f8 h2g1
>>>>file "xboard.cpp", line 855, assertion "mate!=0" failed
>>>>
As for the mate announcements, for the defense of the author, they are not described precisely in the UCI specs. You told me what the correct way was
Fabien.