Calling IPython from a virtualenv
I understand that IPython is not virtualenv-aware and that the most logical solution to this is to install ipython in each virtualenv seperately using
pip install ipython
So far so good. One thing I noticed is that if the system-wide copy of IPython is called from within a virtualenv using $> ipython
before IPython is installed under this virtualenv, subsequent $> ipython
commands will continue to bring up the system-wide ipython copy.
On the other hand, if ipython is not called prior to installing it under a virtualenv $> ipython
will bring up the newly installed copy.
What is the explanation for this?
It also makes me wonder if this behavior means I should expect some trouble down the way?
alias ipy="python -c 'import IPython; IPython.terminal.ipapp.launch_new_instance()'"
This is a great way of always being sure that the ipython instance always belongs to the virtualenv's python version.
This works only on ipython >2.0.
Source
The answer given by @SiddharthaRT is good! Following this approach, it is simpler for me just:
python -m IPython
This will use the module IPython through the python bin, ensuring that it refers to the bin from the virtual env.
You can force IPython to use a virtual environment if available by adding file below to ~/.ipython/profile_default/startups
:
import os
import sys
if 'VIRTUAL_ENV' in os.environ:
py_version = sys.version_info[:2] # formatted as X.Y
py_infix = os.path.join('lib', ('python%d.%d' % py_version))
virtual_site = os.path.join(os.environ.get('VIRTUAL_ENV'), py_infix, 'site-packages')
dist_site = os.path.join('/usr', py_infix, 'dist-packages')
# OPTIONAL: exclude debian-based system distributions sites
sys.path = filter(lambda p: not p.startswith(dist_site), sys.path)
# add virtualenv site
sys.path.insert(0, virtual_site)
I recommend naming it 00-virtualenv.py
so changes will be made as early as possible.
Note: Make sure ipython is installed in the new virtual environment to get this to work.
Activate your virtual environment by using source ~/.virtualenvs/my_venv/bin/activate or by running workon my_venv (Depending on how you've installed the my_venv virtual environment)
Install ipython
pip install ipython
- Now run ipython from my_venv.
If it still loads the system's ipython,then run
hash -r