Installation Bundle


#1

I was speaking to instructors about using more modern tools in our some of our courses (there is a great deal of emphasis placed on MATlab). There are a number of courses that use Python in their courses and primarily cite the reason that Anaconda permits a ‘one-and-done’ installation and students are up an running and everything just works.

I know this is something that can be set up in a docker container or a homebrew recipies, but that is a little beyond new users. The preview build that includes a terminal pane is a great step in this direction whidh provides new users a single IDE for all the things.

The bundling of R and Rstudio into a single installer would definitely help ease newer users into R. Is this something on the radar at Rstudio?


#2

This is not a direct answer to your question, so I hope I’m not hijacking the thread, but we have summarized the getting-started solutions we use at Duke at https://peerj.com/preprints/3181/. It’s a different approach, which requires no installation on the student’s part but resources on the university side.

As a note, as we mention in the paper, students eventually do learn how to install R and RStudio locally, but the goal is to move that to later in the semester and not make it be a first day bottleneck.


#3

Nope, and thank you for the resource! This is similar to what we do in other computing environments on our central VMware servers. It’s easy to have everything the students need already done, but many aren’t comfortable with virtual machines and elect to try to install things on their own. I really like the deferred installation, which is something to try when they have a better sense of the program and greater self-efficacy.

My issue is setting up environments for BYOD workshops as well as installing the same environment on students personal laptops. Time is always tight and we can’t really defer installation. While it looks straightforward, many students are uncomfortable with even installing software not from an app store :expressionless:. Most of the confusion seems to result in the two separate installations and students not recognizing that they are indeed different programs.

Something easily done via a single click and drag package installer would alleviate a lot of problems.


#4

I completely agree with this! Plus, it’s not always straightforward to debug issues that arise at this stage.

I would guess the single click and drag installer would need to have a version selector, unless the default is to use latest versions of both R and RStudio.


#5

I would settle for the current stable versions of each. The choose your own adventure installer, while having its uses, would be more load on already strained self-efficacy and confidence.


#6

Take a look at this post.


#7

This sounds similar to what Ninite does, and python users do have that option.

Anaconda also does quite a few other things behind the scenes to make life easier, such as package installation. R does install packages from a default package list, which can be changed; this might be antithetical to your ends though.

There are some bundling packages as well, such as library(tidyverse). RStudio also has an Install dependencies checkbox :ballot_box_with_check: within the Install Packages window, which will work similarly in some cases.


#8

Thank you for the resource! I thought about a batchfile approach before, but we get a roughly equal split between windows and osx, with a smattering of linux. This is again another layer of complexity for the problem.

@tdawry, most instructors appreciate the bundling of modules in anaconda, and the ability to run jupyter out of the gate. Customizing that list is also a good thing to know, as we can tailor it depending on the workshop/class needs and level of proficiency of the students.