Sometimes getting all the software you need for Scientific analysis updated and running smoothly in OS X can be a bit annoying. For most things, there’s Homebrew (a nice post on setting your Mac for Scientific analysis with Python and homebrew is this one.) And if you’re into GIS, you probably know William Kyngesburye’s QGIS and GRASS packages. Michael Barton also provides GRASS packages, including the last stable version (7.0.2 at this moment). All seems good, but I started to have some problems because I wanted to have my Python stuff (Numpy, Scipy, Matplotlib, etc) updated via homebrew (to keep system python untouched) and then I tried to compile GRASS and QGIS with homebrew as well, but that didn’t worked out so well. The osgeo4mac formulas for ‘brewing’ QGIS are still for version 2.8 (the last stable is 2.12) and there is a ‘tap’ for GRASS 7, but despite compiling ok, I couldn’t get 3D viewing to work. Eventually I managed to get all working. This is how:
Install python, numpy, scipy, matplotlib via homebrew.
Install Michael Barton’s frameworks and GRASS 7.
Install QGIS from William Kyngesburye.
The last part is the tricky one. Homebrew install everything under /usr/local, and this needs to be in your PATH. The usual approach is to add this line to
Doing this will make homebrew’s python the one to be called if you type ‘python’ in a terminal window. But it will also change which python GRASS and QGIS will use. The thing is that both GRASS and QGIS packages are made to use OS X system python, not the one you installed with homebrew. The trick is to set a GRASS_PYTHON variable in .bash_profile, pointing to system python.
Now, the really tricky thing is that GRASS_PYTHON must be called before setting the path for homebrew. So when you start GRASS, it will find the right python. Like this:
export GRASS_PYTHON=/System/Library/Frameworks/Python.framework/Versions/Current/bin/pythonw export PATH=/usr/local/bin:$PATH
This got my setup working fine, with 3D view and all. And if you want to use homebrew’s python for analysis inside GRASS (with pygrass) you just have to call
/usr/local/bin/python from GRASS’ terminal.