[SOLVED] Variable changing

[poed6xDMzRp7]

I do not know how to explain this...:

I have this piece of code:

Code: [Select]

dplaces := $01 shl (4*dplaces)-1;

num1.p2 := num1.p2 and dplaces;
num2.p2 := num2.p2 and dplaces;

num1.p2 := num1.p2+(dplaces+1);
If I am not mistaken, if "dplaces = 2" the first line must result in 255. But when I am debugging the program it keeps changing its value, sometimes when I step into the code it results in 255 then add some code, recompile, debug and it is at different value (32767, 3, etc...).

What am I doing wrong?

Thank you!

[Cyrax]

Do not use any optimization levels when debugging your code. Disable it by giving -O- as parameter to FPC compiler.

[poed6xDMzRp7]

Do not use any optimization levels when debugging your code. Disable it by giving -O- as parameter to FPC compiler.

I already checked for that. Optimization is off, but to clear any doubts I inserted the -O- to the compiler command and the problem persists.

[Cyrax]

Do you initialize dplaces to zero or some other value somewhere in your code?

[poed6xDMzRp7]

Do you initialize dplaces to zero or some other value somewhere in your code?

It is passed as parameter of the "function". When debugging it is at 2 and after the line it is at an unexpected value that changes whenever I add or remove code from the source.

[eny]

Better show us the full code that gives the unexpected result.

[Eugene Loza]

Range check error possible.
try compiling with {$Q+}{$R+}. I always leave those 'on' until the program is finished.

[poed6xDMzRp7]

Range check error possible.
try compiling with {$Q+}{$R+}. I always leave those 'on' until the program is finished.

The problem seems to have stopped after this.

But if I remove the directives... The problem comes back + if I change the variable type it simply crashes with a SIGSEGV. I continue not understanding...

Will I have to compile my final program with those directives?

[eny]

http://www.ebay.com/bhp/magic-crystal-ball

[poed6xDMzRp7]

Better show us the full code that gives the unexpected result.

It seems it was a problem somewhere else because I started a new project with the same code, saved to an empty folder and the problem seems to have vanished.

I consider it solved for now. Thank you!

[eny]

It seems it was a problem somewhere else

It (almost) always is; so showing the full source code is the only way to get to the real problem.
Good that it works now.