Check Extensions

[theturk1234]

Hi all,
I'm having trouble with check extensions. I know it should be ridiculously simple to implement but it's causing a crash.
At each node, if we're in check I increment the depth by one. That's it.
But the engine crashes after a short time thinking.
Has anyone experienced this problem? If so, how did you fix it?
I'm sure I'm doing something really simple the wrong way......

David Cimbalista

[abulmo2]

Hi all,
I'm having trouble with check extensions. I know it should be ridiculously simple to implement but it's causing a crash.
At each node, if we're in check I increment the depth by one. That's it.
But the engine crashes after a short time thinking.
Has anyone experienced this problem? If so, how did you fix it?
I'm sure I'm doing something really simple the wrong way......

When a program crash I use a debugger to understand what happens. It is possible your last addition reveal an older bug that was silent.

[theturk1234]

Hm.....I can try that.
But I'm certainly not an expert at using gdb...

David

[Desperado]

Hi all,
I'm having trouble with check extensions. I know it should be ridiculously simple to implement but it's causing a crash.
At each node, if we're in check I increment the depth by one. That's it.
But the engine crashes after a short time thinking.
Has anyone experienced this problem? If so, how did you fix it?
I'm sure I'm doing something really simple the wrong way......

David Cimbalista

Hi,

well, there are simple things you can check immediately without a debugger.

First, your search code should include a check for maximum depth like

Code: Select all

int search(int depth,...)
{
  if(depth == MAXDEPTH)
  {
    return evalScore()
  }

  // do usual stuff...
}

The next thing you might check is your code for repetition detection.
There are existing "long" checks like: Qa1+,Ke2,Qa2+,Ke1,Qa1+,Ke2

The first suggestion will help to avoid a crash. The second suggestion will avoid that the engine will hang-up because of a very deep tree but running into the mentioned pseudo code.

If these points are ok, you might check if the check extension is triggered as expected and if really produces the right value and the maths works well.

Good luck.

[theturk1234]

Thanks!

[brtzsnr]

Try AddressSanitizer to find your bug: https://gcc.gnu.org/onlinedocs/gcc/Inst ... tions.html

[theturk1234]

Thanks!
I think this might enable me to find the bug!

David Cimbalista