Dependencies for pypi about yeadon HOT 8 CLOSED

Make sure we put examples/misc in (so they can run tests) in MANIFEST.in.

from yeadon.

Should be able to load the examples as well. Also, the PNG's in the docs, etc.

from yeadon.

sudo apt-get install python-vtk

mayaavi (i think pip install mayavi will get all below)
traits
traitsui
pyface
wxpython
configobj

from yeadon.

MayaVi can't install in a virtualenv because it needs the python wrappers to vtk in the virtualenv too. The best bet here is to install a distribution or from the operating system's repos.

from yeadon.

It does not seem reliable to expect dependencies to be installed when performing python setup.py install, so we should just instruct the users to the packages they need to use yeadon, before installing yeadon.

sudo pip install numpy doesn't even work for me; I think it expects that I have some dev packages installed.

To clarify, I'm trying this in a virtual box with a mostly clean version of Ubuntu 12.10.

from yeadon.

So this is a part of the pain of python installation. These are the instructions I think we should say, in this order:

1. Install a scientific python distribution that has all the dependencies, then install yeadon
2. Install the dependencies from your linux distros package manager (show details), then install yeadon
3. Build dependencies from source (pip or manual download), then install yeadon [this is required for virutalenvs, or symlinking to the proper libraries in your distro)

sudo pip install numpy works if you have numpy's build dependencies. But numpy is a dependency. If you have numpy installed then pip will skip the installation of numpy unless you give the -U flag. This is all expected behavior from current python packaging and installation capabilities (not necessarily desired).

from yeadon.

I think those 3 options make sense.

On Sun, Jun 16, 2013 at 11:07 PM, Jason Moore [email protected]:

So this is a part of the pain of python installation. These are the
instructions I think we should say, in this order:

1. Install a scientific python distribution that has all the
   dependencies, then install yeadon
2. Install the dependencies from your linux distros package manager
   (show details), then install yeadon
3. Build dependencies from source (pip or manual download), then
   install yeadon [this is required for virutalenvs, or symlinking to the
   proper libraries in your distro)

`sudo pip install numpy

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/35#issuecomment-19527619
.

from yeadon.

Ok, we just have to write this in the readme and the docs. There isn't much way around it. numpy, VTK and wypython/pyqt are painful to compile so will focus on these options. I have a branch "virtualenv" that was attempting to describe how to build all of those tools for a virtualenv install but it is too much of a pain.

from yeadon.