MS3 1.2.3 source code released

[elaw]

(Edit: huge post deleted because I figured out my problem... multiple brain farts while editing the .ini file. )

[elaw]

Can I ask one more dumb question? This really relates to the C compiler used and not the MS code itself but I'm not quite sure where else to get the answer!

Suppose I write this code:

Code: Select all

int a, b, c, d, y;
a = b = c = d = 10000;
y = (a + b + c + d) / 4;

If the compiler uses 32-bit math, the above will return (in y) the desired result of 10000. If it uses 16-bit math, the (a + b + c + d) calculation will overflow and the result will be a negative number which of course would be bad.

So does anyone know which method the compiler will use? If it's the latter, would changing the calculation to "y = (long)(a + b + c + d) / 4" fix the problem?

[subwoofer]

y = ((long)a+b+c+d)/4 should work, your code is not guaranteed to work since the conversion to long is optionally after the summation.

[jsmcortina]

y = ((long)a+b+c+d)/4 should work, your code is not guaranteed to work since the conversion to long is optionally after the summation.

Yes, the INPUT to an equation must be cast to a large enough variable so that the output will fit. While apparently illogical, this is correct C.

Casting the inputs to a larger type is particularly important for multiplication.

James

[elaw]

Great, thanks!