WB 4.3.15k bug in output window

[Guenther Simon]

Doing some tests with my new quad I noticed that the latest WB release
at least together with Joker 1.117c produces a funny display bug in the
output window regarding calculation of used time(and nps therefore too).

In each PV line of a depth marked with a star by Joker as in:

Code: Select all

515280 <second: 13*     11     1118   10303246  c4d3 d1d3 d5b4 d3b3 g8h8 c1d2 a6a5 a1e1 h7h6 e4c3 e7d6 g5e4 d6e7 b3d1

it calculates time as given in seconds instead of centiseconds, thus the output
for this line was 18:38.00 instead of 00:11.18. (nps therefore /100)

Guenther

P.S. I also noticed that Joker crashed 7 times in 8 games in this test,
but this could be also a compatibility problem with BookThinker and
unrelated to above problem except there will be an overflow problem
at long thinking when displaying factor 100 more?
Currently I am doing a new test and set BookThinker and Inbetween
(for giving the hash param to Joker) to compatibility mode for XP...

[H.G.Muller]

This is non-compliance in Joker. At some point I got notion of the fact that WB thinking output allows you to put an extra arbitrary character behind the depth, and thought this might be a nice way to indicate fail lows / highs or incomplete iterations. (As I connect the latter in my mind with intermediate depth.)

I had not realized that when you do this, WinBoards assumes you are GNU Chess, and triggers a patch to compensate for the fact that GNU Chess reports time in seconds rather than centi-seconds. In later Jokers I now print the * that indicates incomplete iterations in the PV, where it doen't have this side effect. I am not sure if I ever released those versions, though. Perhaps I even only did it for Joker80.

Joker was hacked together in 6 month, and I havent worked on it for more than a year now. I already started writing a new engine, prepared for handling SMP, which would unite Joker and Joker80, as it could handle any board size uptio 10x10. So it is not likely I will ever fix this in the existing Joker 1.x.x versions.

I never saw Joker crash. It very occasionally plays an illegal move, or makes a false illegal-move claim, both in connection with castlings. (Seen onece in Joker80 and once in Joker upto now.)

[Guenther Simon]

This is non-compliance in Joker. At some point I got notion of the fact that WB thinking output allows you to put an extra arbitrary character behind the depth, and thought this might be a nice way to indicate fail lows / highs or incomplete iterations. (As I connect the latter in my mind with intermediate depth.)

I had not realized that when you do this, WinBoards assumes you are GNU Chess, and triggers a patch to compensate for the fact that GNU Chess reports time in seconds rather than centi-seconds. In later Jokers I now print the * that indicates incomplete iterations in the PV, where it doen't have this side effect. I am not sure if I ever released those versions, though. Perhaps I even only did it for Joker80.

Joker was hacked together in 6 month, and I havent worked on it for more than a year now. I already started writing a new engine, prepared for handling SMP, which would unite Joker and Joker80, as it could handle any board size uptio 10x10. So it is not likely I will ever fix this in the existing Joker 1.x.x versions.

I never saw Joker crash. It very occasionally plays an illegal move, or makes a false illegal-move claim, both in connection with castlings. (Seen onece in Joker80 and once in Joker upto now.)

Hi HG,

I want let you know that compatibility mode for BookThinker produced
no more crashes in Vista, thus the second problem seems solved for me.
For the star, which unfortunately triggers an old Gnu patch, I guess
I'll find something too, especially as I already call inbetween for giving
arguments, but still I would welcome a later release with a fix.

Guenther

[H.G.Muller]

If inBetween could simply replace any * in the engine output by a space or nothing, that would do it. Problem is that I did not make the PII compile myself, I am only set up to make gcc compiles.