Ben Dichter

Ben Dichter

Research Software Engineer and Founder of CatalystNeuro.

OSX Jupyter Launcher

1 minute read

Published:

If you use Jupyter on a regular basis, the steps to launch a notebook are probably second nature, but if you take a step back, it involves a lot of prior knowledge. A few times I’ve tried to bring brand new eager programmers into the glorious land of Python and Jupyter, but each time I found that the whole flow was really bogged down by this preamble that is pretty technical. I’ll give them an .ipynb file and then show them how to open it

  1. Open Terminal (What’s Terminal? It looks scary.)
  2. Use cd to navigate to where you want.
  3. Now run this special command… and finally you are in the user-friendly land of Jupyter.

Now of course all of these skills are useful, and necessary eventually, but it really bogs down the first lesson in minutia and inevitably leaves the student feeling a bit overwhelmed. There must be a better way! One solution is to set your student up with Jupyter Hub. They’ll just need to click a link and they’ll be up and running in no time! This is a great solution for a lot of cases, but it requires the instructor to set up a server and the student to have internet access, so this doesn’t fit all cases. “Why can’t I just double-click the notebook?” the student will ask (or be too embarrassed to ask). Well… um… why can’t you? Now you can. Here’s how.

Download me! and double-click to unpack and drag to Applications or where ever you want to keep it.

Navigate to a notebook in Finder, right-click and choose “Get Info”, then expand “Open with:” choose “Other…” from the dropdown menu. Now navigate to and select run_jupyter_notebook. Now select “Change All…”

change notebook settings

Now you can double-click your notebooks to start them!

Caveats

  • This only works on Macs right now (sorry Windows. Linux users, y’all chose this life.)
  • Every time you double-click, it opens a new Terminal window.
  • This won’t run a notebook in a virtual or conda environment.

You can still open notebooks the normal way if you need to have more control over how the notebook is launched.

You May Also Enjoy

The new Code Crafter package for Python AST transformations

2 minute read

Published:

I built a new tool! Code Crafter is a Python library designed for manipulating Python source code through Abstract Syntax Tree (AST) transformations. This tool simplifies the process of programmatically editing Python code, allowing developers to find and modify specific data structures such as lists, dictionaries, and sets within their code.

Stacked Step Plot

1 minute read

Published:

Matplotlib provides a stackplot, which stacks area, and a step plot, which provides steps, but there is no stacked step plot. This is useful when you have some accumulating resource that have different types that begin and end all at once. I personally had a need for this when I wanted to visualize my man-month commitment to my company’s funded projects over the next 7 years.

Grouped Bar Plot

1 minute read

Published:

I often have a need to plot a grouped bar plot. Matplotlib provides this example, which is helpful, but not quite generalizable enough for my needs, as it only shows how to group 2 categories together. Here is a generalization of that tutorial that was very helpful for me and I hope is helpful for others as well.

Git Timesheet

3 minute read

Published:

I recently faced a situation where I needed to assess the amount of work done by each member of a team on a project that has spanned over a year. That project has a git repo, and I could see when each person made a commit. I decided to break it down by weeks. Whenever a person submitted any commit to the repo on any branch, I counted them as working on the project for that week. Of course this is imperfect- someone could work a lot and make no commits for that week and someone could have submitted a commit but might have worked very little. Still, this seems like the most fair way to assess work I could think of.