Teaching julia starting 20 jan. v0.5 or v0.6?

floswald · December 30, 2016, 11:04am

Hi all,

I’m teaching a course using julia. it starts around the 20th of january. based on stefan’s announcement here: 0.6 release timeline I’m uncertain of what the best option is for my students. I’m inclined to run the course on v0.5, but I’m afraid that students will run into trouble some time mid February when they do Pkg.add("DataFrames"), for example. In the best case they will see a bunch of deprecation warnings, but some packages may actually not work, if the maintainer uses v0.6 functionality. Having students check out a previous release of a package is an option that I would like to avoid.

any thoughts on this are helpful. thanks!

tknopp · December 30, 2016, 11:21am

this is what stable releases are for. Just stick with Julia 0.5 and there won’t be any problems

nalimilan · December 30, 2016, 8:48pm

Regarding DataFrames, I would tell students to do Pkg.pin("DataFrames") at the start of the course too. See this previous discussion.

j_verzani · December 30, 2016, 10:07pm

FWIW, I have found it helpful to create a package that students clone that handles this and other package related issues. I have one package that loads and reexports the main packages. It helps avoid frustration for those students who otherwise have issues with the initial start up of a new language.

floswald · December 30, 2016, 10:22pm

thanks for the pointer to this excellent discussion, I couldn’t have formulated it as well as @dmbates. This is a course on computational economics, so the focus is on writing efficient code for computationally intensive algorithms. with the occasional DataFrame here and there. I’ll do the pin trick.