Author: Miguel A. Ballicora
Date: 09:25:00 11/04/02
Go up one level in this thread
On November 03, 2002 at 12:39:51, Robert Hyatt wrote: >On November 03, 2002 at 10:38:20, Uri Blass wrote: > >>I found that there is a bug in the latest version of movei and it seems based on >>page 130 of the book "the practice of programming" that it is a memory >>allocation error. >> >>I see different behaviour in debug configuration and in release configuration. >>I also saw different behaviour when I only added one printf. >> >>My question is what is the fastest way to detect the bug. > >The most likely reason is an uninitialized local variable. It gets allocated >on the stack, and if you call different things, the stack contents will have >different random values for such variables. > >If you have a "lint" program, run it on the source. These things are pretty >good at detecting such stuff. If you are using gcc, or if you can use gcc, >use -O to optimize, and -Wall to turn on all warnings. The -O option turns >on the code to build the dependency graph for each function, showing which >variables are set where and used where, for optimization. But in doing this, >the compiler will spot places where local variables are used before they are >initialized or set. Uri, if you are using the Microsoft compiler turn the level warnings to 3 (at least) and make sure that you do not have any warnings. It is even a good idea (IMHO) not to have any warnings at even level 4, but that might be too much. The latter is debatable. Miguel > > >> >>I have a lot of arrays and adding a code to check if I wrote outside the >>allocated memory in every place is a lot of work. >> >>I think that it may be better if the compiler can do it for me(or maybe the >>compiler can do it and I do not know how it can do it). >> >>Is there a reason that prevent the compiler to check only in debug mode that >>what I put in the arrays make sense before putting the information in the >>arrays? >> >>Note that I save previous versions of movei and I may try to compare number of >>nodes in debug mode and release mode to try to find the latest version that has >>the bug but the problem is that I cannot know that I have not the bug from >>identical output and I can only know that I have a bug when the output is not >>identical. >> >>Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.