[Pyrex] C compile issues when using Numeric (ArrayType) stuff

Francesc Alted falted at openlc.org
Fri Oct 17 20:59:11 CEST 2003


A Divendres 17 Octubre 2003 20:18, Norman Shelley va escriure:
> There were two C files attached.   I assumed one would use "diff" to see
> the lines I removed/modified.

Yeah, but it should be better to attach just the diff so that people can see
it more quickly. From your sources:

$ diff /tmp/numeric_test.c /tmp/numeric_test_fixed.c
< /* Generated by Pyrex 0.9 on Fri Oct 17 10:56:48 2003 */
---
> /* Generated by Pyrex 0.9 on Fri Oct 17 10:43:06 2003 */
341d340
<   ((PyObject*)__pyx_v_Bfortran) = Py_None; 
Py_INCREF(((PyObject*)__pyx_v_Bfortran));
487,488c486
<   Py_DECREF(((PyObject *)__pyx_v_Bfortran));
<   ((PyObject *)__pyx_v_Bfortran) = ((PyObject *)__pyx_3);
---
>   __pyx_v_Bfortran = (PyArrayObject*)__pyx_3;
494c492
<   ((PyObject *)__pyx_v_B) = ((PyObject *)__pyx_v_Bfortran);
---
>   __pyx_v_B = __pyx_v_Bfortran;

First of all, I think that these C files are coming from different Pyrex
sources (see the different header timestamps).

So, I guess (but not entirely sure) that the difference should be restricted
to:

494c492
<   ((PyObject *)__pyx_v_B) = ((PyObject *)__pyx_v_Bfortran);
---
>   __pyx_v_B = __pyx_v_Bfortran;

Much more clear, isn't it? Then, knowing the errors of your HP compiler
would be very useful.

-- 
Francesc Alted





More information about the Pyrex mailing list