The advantages of using MacPorts over EPDFree and other packages are:
The only downside is the long wait for the initial installation (which can take hours) since everything is compiled from source. But this can be left overnight for example, and does not require continuous attention.
The following instructions here are adapted from here
First ensure that XCode version 3 or 4 is installed. To check if XCode is installed type gcc on the command-line. If you get gcc: command not found then XCode is not installed. To install it, use the install DVDs for your Mac, or if you are using MacOS 10.6 or 10.7, you can also get XCode 4 from the App Store.
Download the DMG image for MacPorts from here - be sure to pick the correct one for your MacOS X version. Mount the disk image and run the installer.
Note
if needed, it is possible to install MacPorts without root/admin privileges - see here for more details.
Go to the terminal and update the package index:
sudo port selfupdate
To install Python and most of the basic Python packages, run:
sudo port install py27-matplotlib py27-numpy py27-scipy py27-ipython py27-pip
Note that this will probably take several hours, and is best done overnight.
To install the packages required for this course, use:
sudo port install py27-pyfits py27-pywcs py27-vo py27-asciitable py27-atpy py27-aplpy py27-mayavi py27-beautifulsoup
MacPorts will automatically take care of any dependencies, including py27-pyregion and py27-parsing. You may encounter an error installing py27-mayavi, in this case you can skip installing it. It is not essential for the course.
Note
by default SQL dependencies are not installed for py27-atpy, and the Montage dependency is not installed for py27-aplpy. If you want to install these, use sudo port install py27-atpy +sql) and sudo port install py27-aplpy +montage respectively).
If it does not already exist, create a folder called .matplotlib in your home directory and copy the default matplotlibrc file to it:
cp /opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/matplotlib/mpl-data/matplotlibrc ~/.matplotlib/
Edit .matplotlib/matplotlibrc and change the following line:
backend : Agg
to:
backend : MacOSX
This ensures that when you use Matplotlib interactively, a window will pop up. Optionally, you can uncomment and set the image.origin and image.interpolation lines to:
image.origin : lower
image.interpolation : nearest
These settings are likely to be the most useful for plotting astronomical data.
Make this Python installation the default:
sudo port select --set python python27
sudo port select --set ipython ipython27
Note that this is optional - you can also invoke this Python installation by using python2.7 and ipython-2.7.
To do a test whether you have a functioning core scientific Python installation, do the following to check version numbers. First on the command line check the version numbers of python and ipython:
python -V
ipython --version
Then run ipython from the command line with the --pylab flag:
ipython --pylab
and inside ipython run the following python commands:
import numpy
import scipy
import scipy.linalg
import pylab as plt
from mayavi import mlab
print numpy.__version__
print scipy.__version__
print matplotlib.__version__
x = numpy.linspace(0, 20, 100)
plt.plot(x, sin(x))
print scipy.linalg.eig([[1, 2], [3, 4]])
The commands above should succeed with no errors. The version numbers should meet the requirements, and finally you should see a plot of a sine wave.
To check the other required packages, do the following also from within ipython:
import asciitable
import pyfits
import pywcs
import atpy
import aplpy
If all the above commands run without errors, you’ve installed everything successfully! There is no need to install anything further for the course, but read on for instructions on how to install future python packages you may need using MacPorts.
To check if a package is available through MacPorts, you can do:
$ port search pyfits
py25-pyfits @2.4.0 (python, science)
Python interface to FITS formatted files
py26-pyfits @2.4.0 (python, science)
Python interface to FITS formatted files
py27-pyfits @2.4.0 (python, science)
Python interface to FITS formatted files
Found 3 ports.
Be sure to install the one for the correct Python version (py27-* if you’ve been using the above instructions).
Before installing the package, you can check what variants are available:
$ port variants py27-aplpy
py27-aplpy has the variants:
[+]avm: Include support for AVM meta-data
[+]ds9: Include support for DS9 region files
montage: Include support for Montage reprojection
[+]rgb: Include support for RGB images
universal: Build for multiple architectures
Variants listed with [+] are installed by default. To remove a default variant, use -variant. To include a variant not installed by default, use +variant. For example, the following will install APLpy with support for Montage, but not for RGB images:
sudo port install py27-aplpy +montage -rgb
Warning
Do not use sudo when installing packages not in MacPorts! Read the following instructions carefully to avoid any issues.
When installing packages not in MacPorts, first make sure that you remove your .pydistutils file if you have one, then simply use:
python setup.py install --user
(do not omit --user and do not use sudo). This will place user installed packages in ~/Library/Python/2.7/lib/python/site-packages where they will automatically be picked up by the Python installation, without messing up the MacPorts file structure. Note that you can also install packages in a similar way using easy_install-2.7 and pip-2.7 by specifying the --user option.
The reason for installing packages with --user is that if instead you install packages using sudo python setup.py install, the packages will be installed inside the MacPorts tree, but MacPorts won’t be aware of it, so this could cause issues in future if the package is installed via MacPorts. As a rule of thumb, don’t ever install anything into /opt/local/ other than via the port command. If you did mistakenly install packages to the MacPorts directory, just go to /opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages and remove the files relating to the package you installed.