Setup suggestion - Usage of Google Colab notebooks

[tarunbonu]

Hi there!

Google colaboratory (colab) notebooks are a good way to have everything set up to use instead of having to download and install Python or Anaconda tools locally on the system. This can be a good way to at least get started and be ready for the workshop. Google colaboratory has all the required libraries downloaded and installed without having to sign in with an email id. If it is a requirement to have all the data and notebooks be saved and stored, participants can sign in and colab will save the files in the respective Google drive.

[slel]

In the setup, we could mention several alternatives ways to get started without installing things on one's computer:

• Binder
• CoCalc
• Colab
• Kaggle
• ...

This chart by CoCalc compares some online solutions for Jupyter: https://cocalc.com/doc/compare.html

Personally I always get people started on CoCalc. Among other things, I like that:

• tons of free software comes preinstalled including lots of different Jupyter kernels (C++, Julia, Octave, Python, R, Scala...)
• it's a free software solution (one can install one's own CoCalc server)
• the hosted solution at cocalc.com
  • can be used for free for basic usage
  • has ephemeral mode requiring no account creation
  • offers persistence if one creates an account
• the CoCalc team will usually install software system-wide when one sends a request for widely useful free software that is not already installed there
• the courseware there works very well for workshops, allowing the instructor to send all required files to the learners' CoCalc projects
• the company that runs CoCalc aims at raising money to fund development of free software for science

[ldko]

Hi @tarunbonu and @slel thank you for sharing your thoughts on this issue. Online software solutions is something that comes up regularly, and I have seen individual learners try to use such tools in workshops I have been involved with.

The Carpentries have long preferred learners installing software on their own laptops with the thought they are more likely to continue using the tools that way. Other reasons to support local installations include:

• keeping all learners in a workshop on the same solution to avoid the need for separate instructions and troubleshooting depending on the platform
• lag in typing when using an online system
• spotty Internet connections
• needing to save locally and upload datasets to the cloud
• sessions on cloud based solutions timing out after 10-15 minutes inactivity

However, due to the current situation and needing the option of a software solution for learners who can't get help installing software locally (such as for an online workshop), The Carpentries recently published an article with their position on the matter of working in the cloud during workshops where they recommend MyBinder for Python lessons (when an online solution is necessary) because of its identical interface to JupyterLab. Based on this recommendation, I don't think we will be adding Google Colab to the setup instructions at this time, However, it would be useful to add setup instructions for MyBinder. Would either of you be interested in doing this? Thanks for your time. --Lauren