Hi,
we still support python2.4 I guess.
Not sure what to do there. Maybe, just change it back to something like the
old version for python2.4?
"return Py_BuildValue("(ll)", (long)self->stride0,
(long)self->stride1);"
It's more important we fix the bugs for new pythons anyway.
cu.
On Thu, Jul 28, 2011 at 12:08 AM, Christoph Gohlke <[email protected]> wrote:
>
>
> On 7/27/2011 3:51 PM, Lenard Lindstrom wrote:
>
>> Hi René,
>>
>> No, you have beaten me to this one. I have no access to a 64bit machine
>> for testing. I had a quick look at the patch and it seems sound. I would
>> check the "n" format in return Py_BuildValue("(nn)", self->stride0,
>> self->stride1) though. I don't know when it was introduced to Python.
>>
>> Lenard
>>
>
> Hi,
>
> It was introduced in Python 2.5:
>
> <http://docs.python.org/c-api/**arg.html#Py_BuildValue<http://docs.python.org/c-api/arg.html#Py_BuildValue>
> >
>
> """
> n (int) [Py_ssize_t]
>
> Convert a C Py_ssize_t to a Python integer or long integer.
>
> New in version 2.5.
> """
>
> Christoph
>
>
>
>
>> On 27/07/11 04:45 AM, René Dudfield wrote:
>>
>>> Hi,
>>>
>>> thanks. Looks good to me. You've changed some parts to Py_ssize_t and
>>> c_ssize_t, from the 32bit types.
>>>
>>> Do all the tests pass on 32bit as well?
>>>
>>> I'll try it out later tonight, and apply (unless Lenard beats me to it).
>>>
>>> cheers!
>>>
>>>
>>> On Tue, Jul 26, 2011 at 3:17 AM, Christoph Gohlke <[email protected]
>>> <mailto:[email protected]>> wrote:
>>>
>>> Hello,
>>>
>>> please consider the attached patch. It fixes some array interface
>>> related crashes and test failures on the win-amd64 platform.
>>>
>>> Christoph
>>>
>>>
>>>
>>
>>
