[patch] irrMath.h "isPowOf2"

[REDDemon]

Irrlicht is missing a method for check if a number is a power of 2.

Code: Select all

 
        /** returns if "a" is a power of 2. Behaviour is undefined for floats,doubles and bools*/
        template <class T>
        inline bool isPowOf2(T a)
        {
            return ((a-1) & a) == 0; // 01000 - 00001 = 00111 ... 00111 & 01000 = 0
        }
 

[CuteAlien]

Hm, speed or exactness... to handle zero correct it should be:

Code: Select all

a && ((a-1) & a) == 0;

or (imho less confusing)

Code: Select all

return a && !((a-1) & a);

[hendu]

I thought 0 is a valid pow of 2?

[REDDemon]

Irrlicht is missing a method for check if a number is a power of 2.

Code: Select all

 
        /** returns if "a" is a power of 2. Behaviour is undefined for floats,doubles and bools*/
        template <class T>
        inline bool isPowOf2(T a)
        {
            return ((a-1) & a) == 0; // 01000 - 00001 = 00111 ... 00111 & 01000 = 0
        }
 

yes that class threat both 0 and 1 as power of 2. because in integers number dominion:

2^-inf

is still 0 and a so 0 is a power of 2 (and a power of any other number) and i think is correct. If you prefer to put into irrlicht that 0 is not a power of 2 just document it
but AFAIK with integers is correct that 0 is a power of any number to -inf.

[CuteAlien]

I would say 0 is not a valid power of 2 as negative infinite isn't an integer number, but I guess as long as it's documented what it results in it probably doesn't matter anyway for the cases in which it will be used :-)

[hybrid]

Isn't this just defined on unsigned anyway? So why bother with the template when we can simply add a version taking u32 and avoiding all those cases where a user implements an undefined behavior?

[REDDemon]

yup good Idea.

[gerdb]

hi, i guess a feature request for nextPowerOf2(u32 value) and prevPowerOf2(u32 value) will be advised too.

[REDDemon]

maybe you mean.. "nearestUpperPOT2 or "nearestLowerPOT2"?