PyArt Install

[nguy]

I managed to kill the very old pyArt version Scott installed on my machine a long time ago. So I am reinstalling.

FYI, I have anaconda installed and am working of many of their distributions. The reason for this was simply that NCAR suggest this for the eventual PySolo software. I wanted to give you feedback on my install process and I'll address a couple of things below that I hope are helpful in numeric fashion. Basically as I encounter them or remember.

So you know some package versions:

nickguy[//anaconda]: conda list numpy

packages in environment at //anaconda:

numpy 1.7.1 py27_0
nickguy[//anaconda]: conda list scipy

packages in environment at //anaconda:

scipy 0.12.0 np17py27_0
nickguy[//anaconda]: conda list matplotlib

packages in environment at //anaconda:

matplotlib 1.3.0 np17py27_0
nickguy[//anaconda]: conda list netCDF4

packages in environment at //anaconda:

netcdf4 1.0.5 np17py27_0

1. I had to install baseman using > conda basemap install
   This resulted in matplotlib going down to version 1.2.1
   So I had to manually update matplotlib to v 1.3.0 using conda manager.
2. You might want to note that pyproj is also a needed dependency. I searched and found at:
   http://code.google.com/p/pyproj/

It installed with no issues.

[josephhardinee]

Nick, you might want to just include the pip freeze output if you want to list the packages you are using. It will give you a nice itemized list.
Also I'd include what platform you are on(mac/windows/linux and version). If on a mac, I personally preferred using homebrew and pip for most of the software installs. Linux, package manager + pip.

[nguy]

Hey Joe,
The original cut and paste from an email I was going to send. Didn't work too well! :)
Good ideas! I did fail to mention I'm working on a MacOSX (10.7.5). I generally use homebrew for most packages, though I require some older ones that have not been ported. I'm new to the python world so still feeling it out. I'll check out pip also. As I mentioned, I was sticking with anaconda based on a couple of recommendations.

Anyways, here's the pip freeze output (in completion):
Cython==0.19.1
Flask==0.10.1
Jinja2==2.7.1
MDP==3.3
MarkupSafe==0.18
PIL==1.1.7
PyAudio==0.2.7
PySAL==1.6.0
PyYAML==3.10
Pygments==1.6
SQLAlchemy==0.8.2
Sphinx==1.1.3
VTK==5.10.1
Werkzeug==0.9.4
apptools==4.2.0
astropy==0.2.4
atom==0.3.2
basemap==1.0.6
binstar==0.3.0
biopython==1.61
bitarray==0.8.1
bokeh==0.1.1
boto==2.12.0
casuarius==1.1
chaco==4.3.0
conda==2.0.3
configobj==4.7.2
cubes==0.10.2
distribute==0.6.45
docutils==0.11
enable==4.3.0
enaml==0.7.19
envisage==4.3.0
gevent==0.13.8
gevent-websocket==0.3.6
gevent-zeromq==0.2.2
glpk==0.3
greenlet==0.4.1
grin==1.2.1
h5py==2.2.0
ipython==1.1.0
itsdangerous==0.23
keyring==3.0.1
llvmmath==0.1
llvmpy==0.12.0
lxml==3.2.3
matplotlib==1.3.0
mayavi==4.3.0
meta==development
netCDF4==1.0.5
networkx==1.8.1
nltk==2.0.4
nose==1.3.0
numba==0.10.2
numexpr==2.0.1
numpy==1.7.1
pandas==0.12.0
patsy==0.2.1
pep8==1.4.6
ply==3.4
psutil==1.0.1
py==1.4.14
pyart==0.0.0
pycosat==0.6.0
pycparser==2.09.1
pycrypto==2.6
pycurl==7.19.0
pyface==4.3.0
pyflakes==0.7.3
pyparsing==1.5.6
pyproj==1.9.3
pysam==0.6
pysolo==0.4
pytest==2.3.5
python-dateutil==1.5
pytz==2013b
pyzmq==2.2.0.1
redis==2.7.2
requests==1.2.3
rope==0.9.4
scikit-image==0.8.2
scikit-learn==0.14.1
scipy==0.12.0
six==1.4.1
spyder==2.2.4
statsmodels==0.5.0
sympy==0.7.3
tables==2.4.0
tornado==3.1.1
traits==4.3.0
traitsui==4.3.0
wsgiref==0.1.2
xlrd==0.9.2
xlwt==0.7.5

[jjhelmus]

Technically pyproj is not required to use Py-ART but you do need it if you want to use GridMapDisplay to plot grids, still I'll add it as a optional dependency.

The below is probably more than either of the two of you need, but in case someone looks at this issue later for hints on installing Py-ART or its dependencies:

• There are a number of ways to install a Scientific Python stack and the other modules that Py-ART depends on. I really like Anaconda, Enthough Canopy or EPD for a minimal fuss solution. The full version do have a cost, but all three provide a free version for academics.
• On Mac you can also use MacPorts or Homebrew. You will also likely need to install XCode and the command line utilities.
• With Linux you can typically install many of the required Python modules using the package manager, yum or apt depending on your distrobution.
• Pip works well on all platforms for installing many Python modules after the initial Python stack has been installed. Installing NumPy and SciPy with pip has been known to cause problems and is not recommended.
• Finally SciPy has a nice set of instructions for installing the SciPy stack that might be useful.

[jjhelmus]

Just to be clear do to the formatting of the first post. @nguy did you still have an issue installing Py-ART?

[nguy]

Yeah sorry about that @jjhelmus, the install was successful.

[jjhelmus]

No probelm, happy to have the record for future users to find.

[mvanlierwalq]

Piggybacking on Nick's question, has anyone had success using phase_proc_lp with cylp on Mac OS 10.9? I've installed cylp with Cbc 2.8.6 and 2.8.7 (2.8.5 won't compile on 10.9) and in both cases, I get a malloc error when running phase_proc_lp. Everything works fine on my 10.7 and 10.8 machines.

[jjhelmus]

I was only able to get Cbc and CyLP to work on Mac OS X with the following configuration:

• Python 2.7.6 from Anaconda 1.8.0.
• Cbc 2.8.8 (versions 2.8.6 and 2.8.7 also should work) compiled as follows

./configure 'ADD_CXXFLAGS=-stdlib=libstdc++' 'LDFLAGS=-stdlib=libstdc++'
make
make install

• CyLP 0.7.0 installed from git

I created an CyLP issue pointing out the problems getting CyLP working in OS X 10.9.

For anyone using Anaconda on Mac OS X, I have pre-built binaries of Cbc and CyLP (as well as TRMM RSL, pyproj and Py-ART) in my binstar repository. These can be installed using the conda command, for example to install cylp:

conda install -c http://conda.binstar.org/jjhelmus cylp

[mpy]

Thanks to a hint by @jjhelmus, I was able to upload CyLP binaries to pypi. So now the easiest way to install CyLP on Mavericks is to run "easy_install cylp" (No CBC installation required).

[jjhelmus]

Closing this issue since much of the information is now out of date. CyLP has been updated and builds much more robustly without modifications and pyproj is no longer required in any way by Py-ART.