Description:
------------
This a followup to bug #48455 which has been closed.

I observed same bug with different values.

I'm affraid rasmus' answer is incorrect.

3498 is represented with 12 bits. A float holds 24 bits which is far enough to contain 3498 without any single bit of precision loss.

Anyway, the same program in C works fine:

#include <stdio.h>

main()
{
   float f;

   f=34.98*100;
   printf("%6d %f\n",(int)f,f);
}

thus the problem is NOT with the float representation.

Moreover the following script should return twice the same thing:

$x=34.98*100;

print "x:".$x." sprintf:".sprintf('%d',$x);

but the unexpected result is:

x:3498 sprintf:3497

If the first conversion from float to string works fine, why doesn't the second conversion work ?