[SOLVED] assigning variable to a pointer w/ a type cast

doughyi8u · 08-16-2014, 11:09 AM

This is confusing me. I dont' even know if it's possible but what I'm trying to do is assign to a pointer variable of type float from an int.

Code:

float *p;
int y = 120;

p = (float*)y;
printf("p is %f\n, *p);   /segmentation fault
printf("p is %f\n, p)     /memory location?

return 0;

NevemTeve · 08-16-2014, 12:10 PM

Possible, but makes no sense, causes segmentaion fault. Why would you do that?

Code:

printf("p is %p\n, p);     //memory location
printf("y is at %p\n, &y);     //memory location

return 0;

psionl0 · 08-16-2014, 12:26 PM

According to your code, p is a pointer to a floating variable who's memory location is 120. No wonder you are getting seg faults.

One way to achieve what you want is

Code:

p = malloc(sizeof(float));
*p = (float) y;

doughyi8u · 08-16-2014, 04:34 PM

Code:

p = malloc(sizeof(float));
*p = (float) y;

This doesn't compile
How do you assign to a pointer of type float from a non-pointer variable of type int.

ntubski · 08-16-2014, 05:49 PM

Quote:

Originally Posted by doughyi8u

I'm trying to do is assign to a pointer variable of type float from an int.

Nobody gets up in the morning and says "today I'm going to assign a float pointer from an int". Presumably you're trying to solve an actual problem, why don't you tell us what it is.

doughyi8u · 08-16-2014, 05:52 PM

I'm trying to *fully* understand pointers

psionl0 · 08-16-2014, 09:48 PM

Quote:

Originally Posted by doughyi8u

This doesn't compile
How do you assign to a pointer of type float from a non-pointer variable of type int.

Hmmm, it compiles just fine when I do it. Why don't you show us what you wrote? There is bound to be some typos there.

p = malloc(sizeof(float)); reserves enough memory in the heap to store a floating point number and stores the address of that memory in p.

*p = (float) y; casts (converts) y into a floating point number and stores the result in the memory location pointed to by p.

NevemTeve · 08-17-2014, 12:00 AM

> This doesn't compile

Why don't you try it? If tried, why don't you quote the error message?

mhogomchungu · 08-17-2014, 12:37 AM

Quote:

Originally Posted by doughyi8u

This is confusing me. I dont' even know if it's possible but what I'm trying to do is assign to a pointer variable of type float from an int.

Code:

float *p;
int y = 120;

p = (float*)y;
printf("p is %f\n, *p);   /segmentation fault
printf("p is %f\n, p)     /memory location?

return 0;

p = (float*)y;

The above line reads the value of y and then casts it to (float *) and then writes the result to p.
It is equivalent to:

p = (float*)120;

and i dont think thats what you meant,you probably meant to do:

p = (float*)&y;

The above reads the memory address of variable y and then casts it (float*) and then writes the result to p.
This is probably what you wanted.Use this and it will not segfault anymore.

psionl0 · 08-17-2014, 12:52 AM

Quote:

Originally Posted by mhogomchungu

Use this and it will not segfault anymore.

I'm not sure about that. Is sizeof(float) guaranteed to be the same as sizeof(int) ?

If not then even if the program doesn't segfault, you could be setting yourself up for some awful bugs.

ETA *p would not be y converted to floating format.

mhogomchungu · 08-17-2014, 01:13 AM

Quote:

Originally Posted by psionl0

I'm not sure about that. Is sizeof(float) guaranteed to be the same as sizeof(int) ?

If not then even if the program doesn't segfault, you could be setting yourself up for some awful bugs.

ETA *p would not be y converted to floating format.

It will not segfault and it will work as expected on a 32 bit system since int and float both take 4 bytes.

As far as i know:
float always takes 4 bytes.
int is guaranteed to take atleast 2 bytes but it usually take 4 bytes in most modern 32 and 64 bit systems and 8 bytes in some 64 bit systems.

It modern times and on general purpose systems,its safe to assume int takes atleast 4 bytes.

In a 32 bit system where float take 4 bytes and int take 4 bytes,the program will work as expected.

In a system where the width of the two types are different,*p will give different results but will not cause a crash
because *p will read within the width of variable y unless int takes 2 bytes and i doubt anybody has such a system anymore.

psionl0 · 08-17-2014, 01:37 AM

Quote:

Originally Posted by mhogomchungu

It will not segfault and it will work as expected on a 32 bit system since int and float both take 4 bytes.

Even if they are both 4 bytes, the program will not work as expected - at least, not if you are expecting p to point to a floating representation of the number 120.

Sometimes, one might cast a pointer to a particular data type into a character array if the objective was to manipulate the data bit by bit but I can't think of any valid reason for casting an integer pointer to a float pointer.

mhogomchungu · 08-17-2014, 02:12 AM

Quote:

Originally Posted by psionl0

Even if they are both 4 bytes, the program will not work as expected - at least, not if you are expecting p to point to a floating representation of the number 120.

correct,it will not work as expected since the bit pattern of 120 will be different in an variable of type int and a variable of type float.

Lessons to be learned for the OP.

1. Casting is bad,dont do it and if you must,do it while feeling ashamed of yourself.
2. Getting a handle to an object of one type through a pointer type of another object is a bad idea and de referencing the handle is actually illegal,this link could be a good start in finding out why its illegal[1]

[1] http://cellperformance.beyond3d.com/...-aliasing.html

psionl0 · 08-17-2014, 02:26 AM

Casting from void* is still legal. In fact, it is necessary if you want to do anything with the data.

doughyi8u · 08-17-2014, 09:55 AM

Code:

#include <stdio.h>
#include <stdlib.h>

int
main()
{
    float *p;
    //void *p;
    int y = 120;
    //void *x;

    //x = &y;
    p = malloc(sizeof(float));
    *p = (float)y;
    printf("p is %f\n", *p);

    return 0;
}

compiling this gives the following error:
test1.c:13:26: error: invalid conversion from ‘void*’ to ‘float*’ [-fpermissive]