Fw: solver for kpatience spider is sometimes wrong

[Stephan Kulow]

Am 05.04.2017 um 00:41 schrieb Albert Astals Cid:
>> ideally i would get some introductory help from one of the kpatience/spider
>> programmers eg the latest commit i find in
>> https://phabricator.kde.org/source/kpat/ is from
>> https://phabricator.kde.org/p/mlaurent/ is there a way to contact laurent
>> montel?
> 
> I doubt Laurent has any idea about the spider solver.
> 
> Stephan Kulow is your man for that according to
>   git blame ./patsolve/spidersolver.cpp
> 
> I'm adding him to the CC.
> 
Thanks Albert, I'm not following the list.

The patsolve solvers are quite complex and you should know quite a bit
about graphs and recursion *and* of the spider solitair to dive into it :)

Most "bugs" of the solvers are inherited from optimizations within the
solver to cut off moves that appear pointless to dive into. As Albert
already pointed out - spidersolver.cpp is the file to look at.

Most likely the problem you are facing is in its get_possible_moves
function, where it has to find moves on a given position that aren't
pointless *and* prioritize the move for the A* algorithm to proceed.

Greetings, Stephan