[issue6729] Add support for ssize_t

[Ryan Coyner]

Ryan Coyner <rcoyner at gmail.com> added the comment:

You don't want to do c_size_t = c_void_p because that will prevent type checking. We want c_size_t to be integers; setting it to c_void_p will accept other values. The lines that define c_size_t are doing a sizeof check to determine how many bits the CPU supports, and c_size_t should represent unsigned integers [1].

On a 16-bit machine: c_size_t = c_uint
On a 32-bit machine: c_size_t = c_ulong
On a 64-bit machine: c_size_t = c_ulonglong

Now, ssize_t is like size_t, except that it is signed [2]. So if I am not mistaken, all we have to do is:

if sizeof(c_uint) == sizeof(c_void_p):
    c_size_t = c_uint
    c_ssize_t = c_int
elif sizeof(c_ulong) == sizeof(c_void_p):
    c_size_t = c_ulong
    c_ssize_t = c_long
elif sizeof(c_ulonglong) == sizeof(c_void_p):
    c_size_t = c_ulonglong
    c_ssize_t = c_longlong


Patch attached with documentation and unit test.


[1] - http://www.gnu.org/software/libc/manual/html_node/Important-Data-Types.html
[2] - http://www.gnu.org/software/libc/manual/html_node/I_002fO-Primitives.html

----------
keywords: +patch
nosy: +rcoyner
Added file: http://bugs.python.org/file16405/issue6729.patch

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue6729>
_______________________________________