Alaric needs testers

[Peter Fendrich]

Is my new version really that strong?
I am always too optimistic about my latest creations and need some unbiased testing!

You can select among following profiles:

Code: Select all

  - Strength tester
       - slow time control
       - fast time control

  - Bug killer: I will send you a special bug detector version.
    Results are not interesting - only send the crash log to me.
    Alaric will crash about 1 game out of 100 and I have no clue why.


Thanks for your help!

[Dann Corbit]

For finding the crash, I recommend (for a first pass) Splint for C code:
http://www.splint.org/
and PC-Lint for C++:
http://www.gimpel.com/

This alone will detect most of the problems. It takes a while to get to the point where you can ignore the puffery messages and focus on the ones that mean something though (typically, you will wade through a megabyte of complaints).

Then, for runtime detection of problems BoundsChecker is good, and Valgrind is really excellent for Linux platforms. (We use both of these extensively where I work).

I believe that there is a bounds checking version of GCC, but I have not tried it.

[Peter Fendrich]

Dann,
Thanks for your adivces!

We had this discussion on email some time ago. Remember?

I used another tool, IBM's Rational Rose, that should track memory problems. It didn't give me anything and I don't know if I want to put $250 and spent another 3 days on something that probably will not do better.

The crash occurs very seldom and I have checked every possible table and pointer by hand.
It always happens down in the search, depth >= 8, and it can occur both in the Quiesence and the Search member.
I have seen that from TRY/CATCH blocks.

/Peter

[Dann Corbit]

Does the problem go away if you increase the stack?

Perhaps you have some large automatic data objects.