numpy-discussion

[Numpy-discussion] Testing numpy without doing an installation?

[Francesc A. <fa...@ca...>]

Hi,

Is it possible to test a numpy version directly from the source
directory without having to install it? I mean, if I compile the sources
and try to use the package directly from there, I get unexpected
results. For example:

$ export PYTHONPATH=3D/home/faltet/python.nobackup/numpy/trunk
$ python2.4 -c "import numpy;print numpy.dtype([('col1', '(1,)i4')])"
Running from numpy source directory.
Traceback (most recent call last):
  File "<string>", line 1, in ?
AttributeError: 'module' object has no attribute 'dtype'

It would be nice to have a way of testing a recently built version of
numpy prior to install it.

Thanks,

--=20
>0,0<   Francesc Altet     http://www.carabos.com/
V   V   C=C3=A1rabos Coop. V.   Enjoy Data
 "-"

"Be careful about using the following code --
 I've only proven that it works, I haven't tested it." -- Donald Knuth

Re: [Numpy-discussion] Testing numpy without doing an installation?

[Lisandro D. <da...@gm...>]

On 10/13/06, Francesc Altet <fa...@ca...> wrote:
> Is it possible to test a numpy version directly from the source
> directory without having to install it?

I usually do:

$ python setup.py build
$ python setup.py install --home=3D/tmp
$ export PYTHONPATH=3D/tmp/lib/python

and then

$ python -c 'import numpy; numpy.test()'

and finally,  if all was right,

su -c 'python setup.py install'

or

python setup.py install --home=3D$HOME

--=20
Lisandro Dalc=EDn
---------------
Centro Internacional de M=E9todos Computacionales en Ingenier=EDa (CIMEC)
Instituto de Desarrollo Tecnol=F3gico para la Industria Qu=EDmica (INTEC)
Consejo Nacional de Investigaciones Cient=EDficas y T=E9cnicas (CONICET)
PTLC - G=FCemes 3450, (3000) Santa Fe, Argentina
Tel/Fax: +54-(0)342-451.1594

Re: [Numpy-discussion] Testing numpy without doing an installation?

[Francesc A. <fa...@ca...>]

A Divendres 13 Octubre 2006 22:20, Lisandro Dalcin va escriure:
> On 10/13/06, Francesc Altet <fa...@ca...> wrote:
> > Is it possible to test a numpy version directly from the source
> > directory without having to install it?
>
> I usually do:
>
> $ python setup.py build
> $ python setup.py install --home=3D/tmp
> $ export PYTHONPATH=3D/tmp/lib/python

Thanks for your answer Lisandro, but what I want is to completely avoid an=
=20
installation. The idea is to be able to test the local copy version of nump=
y=20
in the development directory while doing small changes on it. Having to do=
=20
the install step slows down the testing phase when doing small changes.

Cheers,

=2D-=20
>0,0<   Francesc Altet =A0 =A0 http://www.carabos.com/
V   V   C=E1rabos Coop. V. =A0=A0Enjoy Data
 "-"

Re: [Numpy-discussion] Testing numpy without doing an installation?

[Stefan v. d. W. <st...@su...>]

On Tue, Oct 17, 2006 at 10:03:03AM +0200, Francesc Altet wrote:
> A Divendres 13 Octubre 2006 22:20, Lisandro Dalcin va escriure:
> > On 10/13/06, Francesc Altet <fa...@ca...> wrote:
> > > Is it possible to test a numpy version directly from the source
> > > directory without having to install it?
> >
> > I usually do:
> >
> > $ python setup.py build
> > $ python setup.py install --home=3D/tmp
> > $ export PYTHONPATH=3D/tmp/lib/python
>=20
> Thanks for your answer Lisandro, but what I want is to completely avoid=
 an=20
> installation. The idea is to be able to test the local copy version of =
numpy=20
> in the development directory while doing small changes on it. Having to=
 do=20
> the install step slows down the testing phase when doing small
> changes.

It would be great if we could get this to work.  One problem is that
distutils won't build the C-modules in place.  Does anyone know of a
workaround?

At the moment, if you run numpy from the source directory, you see the
message

In [1]: import numpy
Running from numpy source directory.

after which you can't access any of the numpy functions.  This would
be due to this snippet in __init__.py:

try:
    from __config__ import show as show_config
except ImportError:
    show_config =3D None

if show_config is None:
    import sys as _sys
    print >> _sys.stderr, 'Running from numpy source directory.'
    del _sys

If we declare set_package_path and restore_path in __init__.py, we can
wrap all imports in numpy with

set_package_path('../../..') # set path back required depth
from numpy import whatever, you, need
restore_path()

That would take care of things on the python side, at least.

Cheers
St=E9fan

Re: [Numpy-discussion] Testing numpy without doing an installation?

[<pe...@ce...>]

On Tue, 17 Oct 2006, Stefan van der Walt wrote:

> On Tue, Oct 17, 2006 at 10:03:03AM +0200, Francesc Altet wrote:
> > A Divendres 13 Octubre 2006 22:20, Lisandro Dalcin va escriure:
> > > On 10/13/06, Francesc Altet <fa...@ca...> wrote:
> > > > Is it possible to test a numpy version directly from the source
> > > > directory without having to install it?
> > >
> > > I usually do:
> > >
> > > $ python setup.py build
> > > $ python setup.py install --home=/tmp
> > > $ export PYTHONPATH=/tmp/lib/python
> > 
> > Thanks for your answer Lisandro, but what I want is to completely avoid an 
> > installation. The idea is to be able to test the local copy version of numpy 
> > in the development directory while doing small changes on it. Having to do 
> > the install step slows down the testing phase when doing small
> > changes.
> 
> It would be great if we could get this to work.  One problem is that
> distutils won't build the C-modules in place.  Does anyone know of a
> workaround?

Actually numpy.distutils supports in place builds of C modules.
However, its rather difficult to make numpy inplace to work for the
following reasons:
- to build a numpy based C extension, you'll need numpy header files
- numpy header files are generated during numpy build
So it is a chicken-egg problem.

One workaround for testing a numpy subpackage would be installing 
numpy and then doing inplace build in numpy subpackage directory. For
example:
  
  cd svn/numpy
  python setup.py install
  cd numpy/fft
  python setup.py build build_ext --inplace

Another workaround would be to install only numpy.core (that contains
numpy headers) and then doing inplace builds in numpy source directory ---
this, however, requires some setup.py modifications.

IMO, doing inplace builds has side effects that can easily lead to
shooting to a leg. While developing numpy, I would recommend always
installing numpy to some place, setting PYTHONPATH accordingly, write
unittests, and run them for testing. When test_*.py files are set up
properly then you don't need to install everytime you modify unittests,
they can be run inplace.

Pearu