And Now We Are Three

One candidate for that "something else" is quality assurance, i.e., the things we do to ensure that our code actually works. We talked last October about why we don't include testing in the core, but some instructors still do cover testing and debugging using this material from the Hacker Within, these notes from one of last year's WiSE workshops, or our Version 4 notes from five years ago. I think we should try once again to combine this with material on code review to create a half-day lesson on QA for the working scientist. I still believe that lesson will only stick if it includes examples showing how to decide what tests to run for actual scientific problems, and what tolerances to use for "right" answers, but Ian Hawke has put together some IPython Notebooks (which you can view here) to get that ball rolling.

Another candidate for "something else" is how to use Make to create a reproducible paper. Make's syntax is famously opaque, and that lesson moves far too quickly for most of our audience, but again, those are things we can fix.

Writing a usable half-day lesson takes me a week: it's one day per hour of class time to put the first version together, then another couple of days to revise it after its first delivery. Crowdsourcing doesn't help much with the first part, but definitely does with the second. If you're interested in taking a run at either of these, or at some other topic, please get in touch: I'll group interested parties together and get a new lesson repository rolling.