TANGO Control System / Bugs / #612 [pytango][8.0.3] failed to build from source on s390

Frédéric PICCA - 2013-06-04

Just for information this bug will block the migration of pytango AND tango into Debian testing.

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Emmanuel Taurel - 2013-06-06

Category: --> PyTango
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Hello Tiago,

I try to understand the problem, so here is the offending code.
So npy_intp and Py_ssize_t are different on powerpc. I do not know why but
this is a fact.
Now I do not understant the comment about the isImage.
Why do you want to know the coordinates x and y ? can't you just push the data in the buffer the way they are filled in the numpy array.
to my opinion the code for the spectrum should work also for the image but instead of using coordinates you could use the total number of elements of the array.

Did I missed something ?

   if (isImage) {
        // Why not use PyArray_ITER_NEXT() instead of PyArray_ITER_GOTO()?
        // We could do a single while(iter->index < iter->size) instead
        // of the double "for".
        // I did this and it worked in the sense that it went across
        // the correct number of elements but... I did not know the
        // x and y position it corresponded! Yes, 'iter' has a coordinates
        // field, but it was always [0,0], never updated!!
        npy_intp coordinates[2];
        Py_ssize_t &x = coordinates[1];
        Py_ssize_t &y = coordinates[0];
        for (y=0; y < dim_y; ++y) {
            for (x=0; x < dim_x; ++x) {
                PyArray_ITER_GOTO(iter, coordinates);

                PyObject* dataObj = PyArray_GETITEM(array, iter->dataptr);
                const object py_data = object( handle<>( dataObj ) );

                buffer[y*dim_x + x] = extract<TangoScalarType>(py_data);
            }
        }
    } else {
        for (Py_ssize_t x=0; x < dim_x; ++x) {
            PyObject* dataObj = PyArray_GETITEM(array, iter->dataptr);
            const object py_data = object( handle<>( dataObj ) );

            buffer[x] = extract<TangoScalarType>(py_data);

            PyArray_ITER_NEXT(iter);
        }
    }

Cheers

Fred

Tiago Coutinho - 2013-07-09

Hi Frederic,

In fact I did not write the comment. It seems a little strange indeed.
I trust the guy who did it (Ramon Sune) so I prefer to leave it like that and do the fix with a simple cast for now.

Anyway I am thinking of trying to remove all numpy dependencies on C++ level so this code will disapeear in the future (if everything goes well :-)

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Frédéric PICCA - 2013-06-15

proposed patch

can you confirm that it is ok ?

0001-feature-forwarded-simplify-the-numpy-array-extractio.patch

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Tiago Coutinho - 2013-07-09
  
  Hi Frederic,
  
  I fixed it before seeing your proposed patch with a simple data type cast.
  Sorry for that.
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Tiago Coutinho - 2013-07-09

status: open --> open-fixed
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Tiago Coutinho - 2013-07-09

Hello,

Fixed in SVN trunk since revision #23071
Will become visible in next PyTango 8.0.4

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Tiago Coutinho - 2013-08-28

status: open-fixed --> closed

assigned_to: Coutinho --> nobody
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Tiago Coutinho - 2013-11-22

status: closed --> closed-fixed
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

[pytango][8.0.3] failed to build from source on s390

Searches

Help

#612 [pytango][8.0.3] failed to build from source on s390

Discussion