I started building this a few weeks ago as a script for my own use, and decided to make it more public: GitHub - heetbeet/juliawin: A portable Julia for Windows, bundled with VSCode, Pluto, Conda & PyCall, and more.
What do you all think? Is this a useful pursuit or is this largely replicated by Julia Pro? If it’s worth it, I’d consider getting a better website and deployment pipeline going.
6 Likes
Haven’t tested it yet but I think that any attempt to bundle up Julia with Jupyter is worthwhile for teaching. In my experience, 90% of students’ issues are Jupyter (or more generally python) related. If we could avoid those that would be a big improvement!
(Will check it out soon and report back.)
1 Like
An alternative route would be to drop python/jupyter and go for nteract instead. IJulia has instructions.
Although I like the idea of dedicated notebooks, last time I checked nteract was way to much in a beta state for it to be used for undergrad teaching. Has the situation changed? Is someone using it for teaching already?
I had the same impression last time I considered using nteract for teaching (Aug 2019). However, recent versions look more stable.
At this point, the only gripe I have is (a) autocomplete does not work (although there is a PR for that); (b) nteract does not update the kernelspec on saving the notebook (at least not if was earlier created in jupyter), which is a bit of a nuisance.
Cool, I’ll check it out. I think I want to go for a setup where you can pick and choose the tools you want packaged, and also build the launcher exe’s accordingly. Then you can choose jupyter or nteract or both (or none).
1 Like