具有不同环境的 iPython（使用 anaconda）

Question

I have just created a new env with python 3.5 using anaconda (called it python35). My root env points to python 2.7.11. I cant seem to launch ipython with this new env, here is what I did

1. in conda prompt, activate required env: activate python35
2. confirm the version: python --version
3. launch ipython: ipython notebook
4. open a notebook and do: import sys; print (sys.version)

Step #2, returns 3.5 but step #4 always gives me 2.7.11, its like ipython is picking up python version from root env. How do I fix this. Thanks for any help!

This question is related but I have already done what it suggests.

Answer 1

AFAIK, different environments in anaconda (and in venv as well) are activated by prepending env path to $PATH environment variable. It means, that if some file (eg, ipython ) is not found in env path (the first entry of $PATH ), the system searches for it in consequent entries of $PATH and finds it in root environment (that stays in $PATH ). To fix the behavior, you need to install its own copy of ipython to anaconda env:

1. In command prompt, activate the environment: source activate python35 (or simply activate python35 , depending on the OS)
2. While in environment, issue the command conda install ipython-notebook

Answer 2

I could confirm the solution above (basically install notebook in active environment) in my case. Two updates from my side:

1. Since Anaconda 4.4 (?) ipython notebook is now jupyter notebook . So I had to install jupyter instead. I guess, deactivate and activate was required afterwards to get the path variables in notebook updated -> checked by python -c "import sys; print(sys.path)" or corresponding command in notebook.

2. If you aren't using a isolated anaconda environment , you may eventual calling the notebook package from your native OS-Python installation, instead the desired from Anaconda. Similarly, I had trouble with cmake or cxx, when I try to compile in Anaconda Environment - the diffent root folder might found in linux bash with eg "which jupyter"

Answer 3

一个除了安德烈·索伯列夫解决方案，您应该切换到conda install ipython更高python3.x版本和注销环境Ctrl+D或conda deactivate然后重新激活