ID: 39680
User updated by: diefans at googlemail dot com
Reported By: diefans at googlemail dot com
-Status: Bogus
+Status: Open
Bug Type: *Math Functions
Operating System: Linux
PHP Version: 5.2.0
New Comment:
I tested (compiled PHP 5.2.0) the DVAL_TO_LVAL macro (in
Zend/zend_operators.c) from version 5.1.4
and could reproduce expected values.
So the DVAL_TO_LVAL macro from version 5.2.0 performs not equal as in
5.1.4
#define DVAL_TO_LVAL(d, l) (l) = (d) > LONG_MAX ? (unsigned long) (d) :
(long) (d)
please check this issue, if it is intended to be there -
Previous Comments:
------------------------------------------------------------------------
[2006-11-29 12:21:22] diefans at googlemail dot com
So all versions prior 5.2.0 are buggy - right?
------------------------------------------------------------------------
[2006-11-29 11:30:22] [EMAIL PROTECTED]
The logic behind this conversion is quite plain:
if (double > LONG_MAX) {
long = (unsigned long) double;
} else {
long = (long) double;
}
You can try to compile a simple program in C and see that the result is
the same.
------------------------------------------------------------------------
[2006-11-29 10:19:36] diefans at googlemail dot com
Description:
------------
since version 5.2.0 when I cast a negative float smaller than
-2147483648 into an int all bits are lost and the minimum Integer value
is returned. For positive floats larger than the max integer value all
is fine.
It is not possible to perform bitwise calculations with negative
overflowing values since all usable information is lost.
Reproduce code:
---------------
echo (int) -40000000000;
echo (int) 40000000000;
Expected result:
----------------
-1345294336
1345294336
Actual result:
--------------
-2147483648
1345294336
------------------------------------------------------------------------
--
Edit this bug report at http://bugs.php.net/?id=39680&edit=1
