The first R Software Carpentry bootcamp in Cambridge took place on January 7 and 8, 2014. Hosted by the Centre for Mathematical Sciences, the bootcamp was organised by two SSI Fellows, Stephen Eglen and Laurent Gatto. There were 24 participants and a long waiting list of those who wanted to learn R, version control and make.
The bootcamp had a broad coverage of using R, from analysing data to building your own packages and sharing them on GitHub. The three modules on programming, packaging and testing were tied together by the usage and examples from the Camweather package. The teaching material for R was based on the material which Laurent and Stephen have developed for the R courses that they run in Cambridge.
Overall the bootcamp received some very positive feedback from the attendees. In the post-bootcamp questionnaire they were asked to say what they learnt that would be useful for them and here is what they said:
I know now how to use R, Git, GitHub and where to look for info. I've seen examples of make and I think I would be able to generate Makefiles with a bit of patience.
Overview of commands and language syntax and how/where to look up additional information for future use. Knowledge on how to setup repository and deal with new versions and perform central updates. Good mix of lecture and hands-on examples. Etherpad was good as it allowed the course to stay on topic.
The idea behind version control and how to implement it in GitHub was a crucial learning for my programming practise from now on. The presenter make it sound very easy and I will definitely start using that not only for version control in codes but also for shared research and writing papers. I will also lean how to create and check packages in R, in a straightforward way demistifying packing creation. Vignettes and markdown will certainly be very useful for future research practice.
There were several challenges that we faced during the bootcamp. The main issue was the pace of teaching which, for the majority was "just right", but for a few attendees turned out to be "too fast" or "too slow". This is a problem that most bootcamps have. In the questionnaire before the bootcamp we asked the participants about their programming experience. All attendees said they have programmed and most of them selected R as their main programming language. Only three said that they write code once a year or less often. But even for that advanced audience there was definitely too much material squeezed into two days and some of the exercises had to be skipped. The good thing is that they are still available online together with solutions.
The shell-on-Windows-issue also came up in Cambridge. One of the attendees was unalble to work with Make on mysysGit (Git Bash). There were also problems with Git. These problems have been known before and it seems that with every new version of Windows, mysysGit and Cygwin, there may be different solutions.
But despite these struggles, the bootcamp was a good start for Software Carpentry in the UK in 2014, and both Stephen and Laurent are keen to run another event in the future.
See also this news item about the Cambridge R bootcamp on the Software Sustainability Institute's website.