##conda / anaconda installation and usage guide
###Anaconda
Anaconda is a free collection of powerful packages for Python that enables large-scale data management, analysis, and visualization for Business Intelligence, Scientific Analysis, Engineering, Machine Learning, and more. These packages are installed via the command-line tool "conda" (see next section). Anaconda is available for Linux, OS X, and Windows, and works the same way for all of them.
###conda
conda is used to:
- Create and swtich between isolated virtual environments (like virtualenv or PyEnv)
- Download and install (mostly scientific) packages for use with Python (like pip)
The conda package and environment management system allows users to install multiple versions of binary packages (and any required libraries) appropriate for their platform and easily switch between them, as well as easily download updates from an upstream repository. It also works in tandem with pip, so each system is aware of changes made by the other.
conda has a concept of environments which are conceptually similar to virtualenvs, but which use filesystem-level hard links to create entirely self-contained Python runtime layouts. By using the ‘conda’ command line tool, users can easily switch between environments, create environments, and install different versions of libraries and modules into them.
conda does not play nicely with virtualenv. You cannot install Anaconda inside a virtualenv, and I personally suggest disabling virtualenv (or PyEnv) entirely before installing and using Anaconda. You won't need them anymore, anyway. It's time to move on.
The most common way of disabling virtualenv would be to find and comment out any lines looking like this inside your .bashrc or .bash_profile:
Change this:
if [ -f $HOME/.venvburrito/startup.sh ]; then
. $HOME/.venvburrito/startup.sh
fi
To this:
#if [ -f $HOME/.venvburrito/startup.sh ]; then
# . $HOME/.venvburrito/startup.sh
#fi
You must also comment out any virtualenv initialization lines like:
workon my_root_environment
or
source $HOME/.virtualenvs/my_root_environment/bin/activate
If you need help with this you can contact Mario.
###Installation of conda / Anaconda
Installation instructions for Linux, OS X, and Windows are available here:
The default parameters / installation locations are fine.
When creating a new environment under conda, we must specify which packages to install. These can include something as basic as python or as expansive as the full Anaconda suite. We'll create a new environment called "bsw" (for "Best Practices Workshop"). We'll start with a base Python 2.7.6 install:
mario@milou1 ~ $ conda create -n bsw python=2.7.6
Package plan for installation in environment /home/mario/anaconda/envs/bsw:
The following packages will be linked:
package | build
---------------------------|-----------------
openssl-1.0.1c | 0 hard-link
python-2.7.6 | 1 hard-link
readline-6.2 | 2 hard-link
sqlite-3.7.13 | 0 hard-link
system-5.8 | 1 hard-link
tk-8.5.13 | 0 hard-link
zlib-1.2.7 | 0 hard-link
Proceed ([y]/n)? y
Linking packages ...
[ COMPLETE ] |########################################################################################| 100%
#
# To activate this environment, use:
# $ source activate bsw
#
# To deactivate this environment, use:
# $ source deactivate
#
mario@milou1 ~ $
Let's check to see which environments I have now:
mario@milou1 ~ $ conda info -e
# conda environments:
#
bsw /home/mario/anaconda/envs/bsw
root * /home/mario/anaconda
mario@milou1 ~ $
The asterisk indicates that I'm currently in the "root" environment."
Next, I'll enter the "bsw" environment:
mario@milou1 ~ $ source activate bsw
prepending /home/mario/anaconda/envs/bsw/bin to PATH
(bsw)mario@milou1 ~ $
Now -- I'm fairly sure we'll be using iPython during the workshop. I'll see which versions are available:
(bsw)mario@milou1 ~ $ conda search ipython
ipython 0.13 py26_0 defaults
0.13 py27_0 defaults
0.13 py26_1 defaults
0.13 py27_1 defaults
0.13 py33_1 defaults
0.13.1 py26_1 defaults
0.13.1 py27_1 defaults
0.13.1 py33_1 defaults
0.13.2 py26_0 defaults
0.13.2 py27_0 defaults
0.13.2 py33_0 defaults
1.0.0 py26_0 defaults
1.0.0 py27_0 defaults
1.0.0 py33_0 defaults
1.1.0 py26_0 defaults
* 1.1.0 py27_0 defaults
1.1.0 py33_0 defaults
ipython-notebook 0.13.2 py27_0 defaults
1.0.0 py27_0 defaults
1.1.0 py26_0 defaults
* 1.1.0 py27_0 defaults
1.1.0 py33_0 defaults
1.1.0 py26_1 defaults
1.1.0 py27_1 defaults
1.1.0 py33_1 defaults
ipython-qtconsole 0.13.2 py27_0 defaults
1.0.0 py27_0 defaults
* 1.1.0 py27_0 defaults
1.1.0 py27_1 defaults
We see the default version is 1.1.0 for Python v2.7. That's fine, so we can install it without specifying a version number:
(bsw)mario@milou1 ~ $ conda install ipython
Package plan for installation in environment /home/mario/anaconda/envs/bsw:
The following packages will be linked:
package | build
---------------------------|-----------------
ipython-1.1.0 | py27_0 hard-link
Proceed ([y]/n)? y
Linking packages ...
[ COMPLETE ] |########################################################################################| 100%
(bsw)mario@milou1 ~ $
Let's see which packages we have installed now:
(bsw)mario@milou1 ~ $ conda list
# packages in environment at /home/mario/anaconda/envs/bsw:
#
ipython 1.1.0 py27_0
openssl 1.0.1c 0
python 2.7.6 1
readline 6.2 2
sqlite 3.7.13 0
system 5.8 1
tk 8.5.13 0
zlib 1.2.7 0
Alright so I'm set for the workshop. Now it's time to go back to my real passion in life, backend web development using the Python-based Django framework.
Somehow, web developers got ahead of the curve and have already begun migrating to Python 3. I think it's reflective of their reckless attitude toward the establishment. Or maybe they're just too iterative to have any legacy code to support. In any event, we'll create a new environment with Python 3 and Django.
First, we'll leave our current environment:
(bsw)mario@milou1 ~ $ source deactivate
discarding /home/mario/anaconda/envs/bsw/bin from PATH
mario@milou1 ~ $
Now we'll see what versions of Python are available to us:
mario@milou1 ~ $ conda search python
... ... ...
python 2.6.8 1 defaults
2.6.8 2 defaults
2.6.8 3 defaults
2.6.8 4 defaults
2.6.8 5 defaults
2.6.8 6 defaults
2.6.8 7 defaults
2.6.9 0 defaults
2.7.3 2 defaults
2.7.3 3 defaults
2.7.3 4 defaults
2.7.3 5 defaults
2.7.3 6 defaults
2.7.3 7 defaults
2.7.4 0 defaults
2.7.5 0 defaults
2.7.5 1 defaults
2.7.5 2 defaults
2.7.5 3 defaults
2.7.6 0 defaults
* 2.7.6 1 defaults
3.3.0 pro0 defaults
3.3.0 pro1 defaults
3.3.0 2 defaults
3.3.0 3 defaults
3.3.0 4 defaults
3.3.1 0 defaults
3.3.2 0 defaults
3.3.2 1 defaults
3.3.3 0 defaults
3.3.4 0 defaults
3.3.5 0 defaults
... ... ...
mario@milou1 ~ $
Plenty of versions available (if you do this search yourself, you'll also see many other packages I'm omitting here, including iPython, Biopython, and more). Let's install python 3.3.4:
mario@milou1 ~ $ conda create -n django-webdev python=3.3.5
mario@milou2 ~ $ conda create -n django-webdev python=3.3.5
Fetching package metadata: ..
Solving package specifications: ...
Package plan for installation in environment /home/mario/anaconda/envs/django-webdev:
The following packages will be downloaded:
package | build
---------------------------|-----------------
python-3.3.5 | 0 18.6 MB
The following packages will be linked:
package | build
---------------------------|-----------------
openssl-1.0.1c | 0 hard-link
python-3.3.5 | 0 hard-link
readline-6.2 | 2 hard-link
sqlite-3.7.13 | 0 hard-link
system-5.8 | 1 hard-link
tk-8.5.13 | 0 hard-link
zlib-1.2.7 | 0 hard-link
Proceed ([y]/n)? y
Fetching packages ...
python-3.3.5-0.tar.bz2 100% |###############################################################| Time: 0:00:08 2.21 MB/s
Extracting packages ...
[ COMPLETE ] |########################################################################################| 100%
Linking packages ...
[ COMPLETE ] |########################################################################################| 100%
#
# To activate this environment, use:
# $ source activate django-webdev
#
# To deactivate this environment, use:
# $ source deactivate
#
Now we need to install Django. Let's enter the environment and see if the Anaconda repos have it:
mario@milou1 ~ $ source activate django-webdev
prepending /home/mario/anaconda/envs/django-webdev/bin to PATH
(django-webdev)mario@milou1 ~ $ conda search django
(django-webdev)mario@milou1 ~ $
Nope! That makes sense of course because it's not a scientific package. We can just download it using pip instead, because both pip and conda make use of Python's setuptools distribution management system.
Important note!!
conda does not by default install pip or setup_tools (though these are included in the anaconda metapackage) so if you need to use pip you must explicitly install it using conda. It may appear that you can use it without installing it, but I assure you that this is merely your computer deceiving you while trying to be helpful: if you examine the pip output closely you will observe that it is installing packages into a location that is not used by your new environment, and a quick which pip will show you that this is because it is a binary not installed by, and unrelated to, this environment.
(django-webdev)mario@milou2 ~ $ conda install pip
Fetching package metadata: ..
Solving package specifications: ...
Package plan for installation in environment /home/mario/anaconda/envs/django-webdev:
The following packages will be downloaded:
package | build
---------------------------|-----------------
pip-1.5.4 | py33_0 1.5 MB
setuptools-3.1 | py33_0 474 KB
------------------------------------------------------------
Total: 2.0 MB
The following packages will be linked:
package | build
---------------------------|-----------------
pip-1.5.4 | py33_0 hard-link
setuptools-3.1 | py33_0 hard-link
Proceed ([y]/n)? y
Fetching packages ...
pip-1.5.4-py33_0.tar.bz2 100% |#############################################################| Time: 0:00:02 712.10 kB/s
setuptools-3.1-py33_0.tar.bz2 100% |########################################################| Time: 0:00:01 421.00 kB/s
Extracting packages ...
[ COMPLETE ] |#########################################################################################| 100%
Linking packages ...
[ COMPLETE ] |#########################################################################################| 100%
Great! Now we can use pip.
(django-webdev)mario@milou1 ~ $ pip install django
Downloading/unpacking django
Downloading Django-1.6.2-py2.py3-none-any.whl (6.7MB): 6.7MB downloaded
Installing collected packages: django
Successfully installed django
Cleaning up...
(django-webdev)mario@milou1 ~ $
Alright so... that's it. That's all I got.