[Checkins] SVN: persistent/trunk/ Worked around test failure due to overflow to long on 32-bit systems.

[Jim Fulton]

On Sun, Aug 26, 2012 at 2:30 PM, Tres Seaver <cvs-admin at zope.org> wrote:
> Log message for revision 127583:

...

> Modified: persistent/trunk/persistent/cPersistence.c
> ===================================================================

...

> @@ -1118,6 +1118,14 @@
>  {
>    if (v)
>      {
> +      if (PyLong_Check(v))
> +      {
> +          long long llv = PyInt_AsLongLong(v);
> +          if (llv > sys_maxint)
> +          {
> +             v = sys_maxint;  /* borrow reference */
> +          }
> +      }

This is wrong.  You're comparing a long long and a PyObject*.

...

>
> Modified: persistent/trunk/persistent/tests/test_persistence.py
> ===================================================================
> --- persistent/trunk/persistent/tests/test_persistence.py       2012-08-26 16:19:07 UTC (rev 127582)
> +++ persistent/trunk/persistent/tests/test_persistence.py       2012-08-26 18:30:20 UTC (rev 127583)
> @@ -504,9 +504,18 @@
>
>      def test_assign_p_estimated_size_bigger(self):
>          inst = self._makeOne()
> -        inst._p_estimated_size = 1073741697 * 1024
> +        inst._p_estimated_size = 1073741697 * 4 #still <= 32 bits
>          self.assertEqual(inst._p_estimated_size, 16777215 * 64)

This test is meant to test that the code that prevents us from
overflowing the 30-bits reserved for storring estimated sizes.  Using
* 2 (rather than * 4) should be enough.

We don't really need to support Python longs, do we?  We just need to
fix the test to not provoke a long.

If someone sets a size >2G on a 32-bit platform, they'll get a type
error.  I don't think this is a big problem.  If they're doing this,
they probably have other problems. :)

I propose to revert this change and change 1024 to 2.

Jim

--
Jim Fulton
http://www.linkedin.com/in/jimfulton
Jerky is better than bacon! http://zo.pe/Kqm