Check out JupyterLab

[NorthGuard]

Can you collaborate? Can you work at the same time in the same notebook? And is that even a good idea?

[SkafteNicki]

Short summary of what I have read:

JupyterLab is cannot be used for collaboration. In time it will completely replace jupyter notebook (so it is really just jupyter notebook 2.0) and it is looking very promising. It has a cleaner look->easier to use, have more options ect. It can easily be installed with pip install jupyterlab. I intend to switch from notebook to lab.

Then there is JupyterHub which is can be used for multi-user collaboration. It should allow multiple of user to interact with the jupyter notebook/lab server at the same time. It is harder to setup, so if we choose to go down this path we probably need to write our own setup documentation for the students. The biggest downside is that it does not support windows (and probably will not in the near future). Therefore i do not consider this a valid option, since many of the students will probably have windows installed on their laptop when they start.

I will continue to look for another option that allow multi-user collaboration.

[SkafteNicki]

I found the perfect tool for collaboration on notebooks:

https://colab.research.google.com/notebooks/welcome.ipynb

I really think that the student will like this form of online collaboration (especially if you keep in mind how many are using sharelatex at this point).

Advantages:

• Free
• Real time collaboration for multiple users
• Works perfectly with google drive
• Can run locally and on hosted google served
• The hosted server already have tensorflow, keras, opencv ect. installed
• The hosted server has GPU support

Only disadvantage:

• It is a google research project, meaning that they can take it away at any point. But google is usually nice enough to let people know some time in advance, so they have the possibility to find other options.
• Only support for .ipynb, not native .py files. Not really a problem since .ipynb is just .py with markdown

The only question we need to answer is how it fits into our framework, but it seems manageable (https://stackoverflow.com/questions/48905127/importing-py-files-in-google-colab).

Closing this for now.