Make Things

Fork Us on GitHub

If you are interested in developing the Software Carpentry content, we have put our materials into these repositories on GitHub.

These older repositories are still accessible, but their content is no longer being maintained:

• website: the old website.
• boot-camps: used for lesson materials and bootcamp home pages Sept. 2012–July 2013.
• lessons_3_0: Version 3 lesson materials.
• lessons_4_0: Version 4 lesson materials.

Pull requests are always welcome, or you can mail us to talk about what you'd like to do and how we can help.

New Exercises, Lessons, and Examples

We welcome new exercises and lessons (particularly lessons that show how the things we teach fit together by working through medium-sized problems). A good exercise has a single clear focus, and should be doable in five minutes or less; for examples, take a look at the "Challenges" sections of our introduction to the Unix shell. We'd particularly like Python exercises that use ipythonblocks and skimage.novice in the IPython Notebook, exercises for our web programming lesson, and ones for using Git and GitHub.

For an example of what we'd like in a longer lesson, please look at the invasion percolation example, which covers topics as diverse as restructuring code to make it more testable, performance profiling, and the advantages of lazy data structures. Such examples can be specific to different research disciplines, which often makes them the most convincing part of our bootcamp. If you're looking for starting points, we'd like to rewrite our examples of constructing phylogenetic trees, counting stars, and steganography. We'd like to add a simple N-body simulation that included discussion of testing in the face of floating-point roundoff, calculating and visualizing spatial correlations between different data sets, something like Prof. Lorena Barba's 12 Steps to Navier Stokes (but shorter, since we only have a couple of hours), and some more gene sequence wrangling. If you'd like to tackle one of these, or have ideas of your own, please email us: we'd be happy to chat.

Related Projects

If you want to help us by programming, there are lots of things we need. Please get in touch if you'd like to tackle any of these, or have projects of your own you'd like to suggest.

Website Makeover

Our website desperately needs a makeover, which requires graphic design talent and CSS skills more than it does Python coding.

Blocks and Images

ipythonblocks and skimage.novice have let us adopt a media-first approach to teaching programming. Enhancements to either of these are very welcome.

Diffing and Merging

The most important project to us is actually a tool that will diff and merge notebooks so that we can work with them more easily in version control. A groups of students are working on this in Montreal as a course project, so it's difficult for other people to help code right now, but we hope to open it up soon.

Badging

We issue badges for instructors, organizers, and people who write lesson materials, and plan to start issuing them soon for people who've taken part in our bootcamps. However, our badging infrastructure has fallen behind updates to the standard, so we need to upgrade it, and to convert our existing badges to the newer format.

Browsercast

Over the summer, David Wolever built a prototype of a tool called Browsercast, which is meant to be a replacement for screencasting. Instead of taking what's happening on your screen and turning it into a video that is large and opaque (i.e., you can't search for text in it or copy/paste things from it), Browsercast replays an IPython Notebook session in the browser in sync with a voiceover. What we need is a plugin for nbconvert (the notebook-to-whatever conversion tool used to generate static HTML pages, PDFs, and so on from IPython Notebooks) that will take the notebook and the annotations saying, "The start of this cell corresponds to time 2:07 in the audio," and produce an HTML page that people can view without having the notebook running on their machine. A couple of volunteers have started looking at this, and more help would be welcome.

Variable Expansion in the IPython Notebook

I've been using http://regexpal.com to teach regular expressions for a while, and then I thought, "Hey, I can do this in the notebook." It urns out that writing extensions (called "magics") for the notebook is pretty easy, so I built one that takes a cell whose first line is a regex, and other lines are text, and shows the matches. Except there's a problem: right now, the notebook interprets the text {something} to mean, "Expand the variable 'something' before passing the text to the extension," so a regex like \d{2-4} (which is supposed to match 2, 3, or 4 digits) doesn't work. We can get around this by doubling up the braces, but the IPython notebook developers think that the right solution is to refactor the code that handles magics so that people can get the original (unexpanded) text. Turns out that's easier said than done, but if you'd like to dive into the guts of the notebook, this is your chance.

Turtle Graphics

It would be really cool to be able to use Python's turtle graphics module for teaching in the notebook. Turtle graphics goes all the way back to the original work at MIT on teaching kids to program and Paper's wonderful book Mindstorms. It's great for teaching, but only if you can see the picture evolve on the screen in front of you: if all you get is the final image, you don't know where to start debugging. Unfortunately, the notebook is set up mostly for batch processing: you press "shift+enter" in a cell, the text is sent to the back-end Python process, the result is sent back and displayed. It should be possible to plug something into the turtle module so that instead of actually drawing, it produces a list of atomic line-drawing commands. These can be sent back to the notebook, where a bit of Javascript can replay the drawing in the browser to give the desired insight.

A Perl Back End

The IPython Notebook doesn't have to use Python as the back-end process: they've defined a JSON-based protocol for communicating with arbitrary processes, and people have already built a replacement back-end in Julia (and are working on one in R). There are lots of Perl programmers in science; a Perl back-end would probably get a lot of love.

Data Structure Visualization

Philip Guo built an online Python tutor that draws what's in memory while the program executes. He managed to get it partially integrated into an earlier version of the IPython Notebook, but it was all duct tape and prayer. It would be very cool to have a notebook extension that would let you say, "Show me the state of memory right now," and produce a static image in the notebook showing what things looked like at that moment. The introspection on the Python side isn't particularly hard—certainly not for basic data types—and if you could make it work for one language, you could re-use the browser-side stuff for other languages.

Editing Diagrams

It would be very cool if we could edit diagrams directly in the notebook. I think this can be done by integrating svg-edit; if it can, it would be useful for people who want to do slideshows using the notebook as well as for teaching.