Before answering that, let me ask you, if it were valid, what would you do with such a pointer?
Suppose you have (inside a function)
float a;
int *ptr; ptr = &a;
*p = 1;
++*p;
return *p;
In such a case, there is no reason at all not to just use an int
variable if it's an int
you want.
Suppose you have (inside a function)
float a;
int *ptr; ptr = &a;
a = 3.14;
++*p;
return a;
In this sort of object aliasing, there can be a use, but allowing constructions like this is a pain for compilers. Compilers are free to assume, and some do assume, that the modification of *p
has no effect on the value of a
, so still return 3.14 unmodified;
If the latter case is something you're looking for, you can use union
, which both makes your code clearer to other readers, and makes your code clearer to the compiler:
union {
float f;
int i;
} u;
u.f = 3.14;
++u.i;
return u.f;
So, to answer: it's to prevent you from shooting yourself in the foot. It's not allowed because there is no way such a pointer is useful. The one case where it could be useful is actually a case where it doesn't work, and there is another language construct that does handle it, and handles it correctly.