pip install fails with missing numpy

[zjttoefs]

There is a obvious workaround.
I'm playing with some fresh virtual machines and would prefer nexpy handling at least the mandatory dependencies.

[rayosborn]

Is it that pip is unable to install numpy when it encounters the dependency?

[zjttoefs]

It does not appear to make any attempt at installing a dependency:

---

$ pip install nexpy
Downloading/unpacking nexpy
Downloading NeXpy-0.5.4.tar.gz (1.5MB): 1.5MB downloaded
Running setup.py (path:/tmp/pip-build-8hOZDz/nexpy/setup.py) egg_info for package nexpy
Traceback (most recent call last):
File "", line 17, in
File "/tmp/pip-build-8hOZDz/nexpy/setup.py", line 14, in
pkg_resources.require('numpy')
File "/usr/lib/python2.7/dist-packages/pkg_resources.py", line 745, in require
needed = self.resolve(parse_requirements(requirements))
File "/usr/lib/python2.7/dist-packages/pkg_resources.py", line 639, in resolve
raise DistributionNotFound(req)
pkg_resources.DistributionNotFound: numpy
Complete output from command python setup.py egg_info:
Traceback (most recent call last):

File "", line 17, in

File "/tmp/pip-build-8hOZDz/nexpy/setup.py", line 14, in

pkg_resources.require('numpy')

File "/usr/lib/python2.7/dist-packages/pkg_resources.py", line 745, in require

needed = self.resolve(parse_requirements(requirements))

File "/usr/lib/python2.7/dist-packages/pkg_resources.py", line 639, in resolve

raise DistributionNotFound(req)

pkg_resources.DistributionNotFound: numpy

---

Cleaning up...
Command python setup.py egg_info failed with error code 1 in /tmp/pip-build-8hOZDz/nexpy
Storing debug log for failure in /home/vagrant/.pip/pip.log
$

---

[prjemian]

How can this be a failure of NexPy?
On Jun 19, 2015 8:11 AM, "Tobias Richter" [email protected] wrote:

It does not appear to make any attempt at installing a dependency:

$ pip install nexpy
Downloading/unpacking nexpy
Downloading NeXpy-0.5.4.tar.gz (1.5MB): 1.5MB downloaded
Running setup.py (path:/tmp/pip-build-8hOZDz/nexpy/setup.py) egg_info for
package nexpy
Traceback (most recent call last):
File "", line 17, in
File "/tmp/pip-build-8hOZDz/nexpy/setup.py", line 14, in
pkg_resources.require('numpy')
File "/usr/lib/python2.7/dist-packages/pkg_resources.py", line 745, in
require
needed = self.resolve(parse_requirements(requirements))
File "/usr/lib/python2.7/dist-packages/pkg_resources.py", line 639, in
resolve
raise DistributionNotFound(req)
pkg_resources.DistributionNotFound: numpy
Complete output from command python setup.py egg_info:
Traceback (most recent call last):

File "", line 17, in

File "/tmp/pip-build-8hOZDz/nexpy/setup.py", line 14, in

pkg_resources.require('numpy')

File "/usr/lib/python2.7/dist-packages/pkg_resources.py", line 745, in
require

needed = self.resolve(parse_requirements(requirements))

File "/usr/lib/python2.7/dist-packages/pkg_resources.py", line 639, in
resolve

raise DistributionNotFound(req)

pkg_resources.DistributionNotFound: numpy

Cleaning up...
Command python setup.py egg_info failed with error code 1 in
/tmp/pip-build-8hOZDz/nexpy
Storing debug log for failure in /home/vagrant/.pip/pip.log

$

—
Reply to this email directly or view it on GitHub
#60 (comment).

[danielballan]

Writing a bulletproof setup.py is always messy and virtually impossible. Has anyone packaged nexpy for conda? That might be the easiest way to get running quickly on virtual machines. If there is interest I can take a look at this.

[rayosborn]

I would be delighted if you were able to get a conda package set up. The popularity of conda was one of the most important reasons for making NeXpy compatible with PyQt, so it's on my to-do list. Not sure why pip wouldn't just install numpy though, since it is clearly listed as a dependency.

[danielballan]

My understanding is that installation-time dependencies (as opposed to runtime dependencies) cannot be automatically installed by pip/setuptools. This is -- if I understand correctly -- a limitation of setuptools. The best you can do is fail informatively, which nexpy is doing.

[tacaswell]

There are ways to setup tools to cleanly install other things first, matplotlib does this, but it requires cajoling. See https://pythonhosted.org/setuptools/setuptools.html#new-and-changed-setup-keywords and setup_requires.

[zjttoefs]

Just adding a setup_requires didn't do the trick. I made the change on a branch and tried:

pip install git+git://github.com/nexpy/nexpy.git@zjttoefs-issue-60

It still fails on the same line. I must admit I have a very bare understanding of setuptools but I don't quite get why the setup needs (imports) numpy (and nexpy).

[tacaswell]

@zjttoefs Your not alone 😉

[zjttoefs]

Thanks @tacaswell !
Okay, I understand those imports now. Removing them requires some extra work, that is neatly avoided by using them.
Maybe stealing a bit from matplotlib would be good.

[rayosborn]

I'm all in favor of raiding matplotlib whenever possible. I won't have time to look into this for a week or so, but will be happy to consider pull requests. By the way, I had to remove PySide from the dependency list in order to allow PyQt4 or 5-only installations. I don't know if there is any way of handling these kinds of dependencies better.

[tacaswell]

As a warning, mpl's setup.py grew organically through multiple generations
of setup tools and has a lot of cruft (that at least I don't understand
well enough to be willing to touch).

h5py might be a better template.

On Fri, Jun 19, 2015 at 11:08 AM Ray Osborn [email protected]
wrote:

I'm all in favor of raiding matplotlib whenever possible. I won't have
time to look into this for a week or so, but will be happy to consider pull
requests. By the way, I had to remove PySide from the dependency list in
order to allow PyQt4 or 5-only installations. I don't know if there is any
way of handling these kinds of dependencies better.

—
Reply to this email directly or view it on GitHub
#60 (comment).