Is there a way to do this? Start an ipython / jupyter notebook inside of a specific virtualenv?
Background: I am taking a Udacity deep learning class that uses the google tensorflow library. The class provides some ipython notebooks as projects / homework to work from.
I have created a python 2.7 virtualenv ("tensorflow") and have installed ipython, jupyter and tensorflow there. I have uploaded the Udacity class ipython notebook into my home directory.
A) From inside the pythonanywhere virtual environment, I can start a notebook on a localhost, but I don't have access to that localhost port, so that doesn't seem to give me access to the running notebook.
B) I can create an empty python 2.7 notebook, and I could cut and paste the Udacity notebook material into that notebook as a workaround (I guess) but the python 2.7 notebook environment does not have access to the tensorflow library in the virtualenv.
C) I guess I could install the tensorflow library into my default (python 2.7) workspace without a virtual environment, but there appear to be many dependencies and this kind of defeats the purpose of having virtual environments. Maybe this would work, but right now I don't want to risk introducing some other package/module dependency.
D) From the files tab, I can select and open one of the ipython notebooks in my home directory, and while this opens with python 3.x, I can go into the kernel tab and select the 2.7 kernel environment. I think this is giving me my default 2.7 home env python, and there is no tensorflow library there.
Other ideas or suggestions? After I wrote this I realized options B) and D) are basically the same.
Thanks.