And we're off! RStudio internship week 1, complete.

First week of internship is over and I can't believe this is actually happening

By Daniel Chen

June 10, 2019

I’m still pinching myself about being one of the RStudio interns this year. It’s an unbelievable opportunity and I’ve been half panicked and fighting imposter syndrome since the announcement was made in March. My meeting with Greg Wilson on Friday (2019-06-07) went something like this:

Greg Wilson: How's the internship going?
Me: I'm panicked, but really excited.
Greg Wilson: Good. That's how interns should feel.

I’m working on the grader package (with Garrett Grolemund and Barret Schloerke) which aims to check code against a solution. This package integrates with the learnr package, that allows an RMarkdown document to be run as an interactive tutorial (more on this after the break).

My internship began with a broken pull request that needs to be fixed.

The learnr/grader packages

learnr allows users to write interactive tutorials/lessons (using RMarkdown) that give instructors to write interactive tutorials (using shiny). Learners will be able to follow the tutorial write and execute R code that will be graded. Under the hood, learnr is an RMarkdown document with runtime: shiny. This allows the user (e.g., instructor) to write tutorials using markdown and provide solutions using code chunks and all the magic of capturing and checking the answer with shiny and grader.

There are two main ways code can be graded.

  • Comparing results
  • Comparing the provided student code with the solution code (i.e., strict checking).
    • One way this can be done is by looking at the abstract syntax tree.
    • Another way would be some fancy set of string matching / regular expressions.

When I started, there were some API changes that were made and I’ve spent my first week going down various rabbit holes fixing tests and getting R CMD check to pass. Barret has been super helpful and supportive by walking me through the code base and helping me fix the broken things. I’m pretty sure he did all the work in my PR.

General goals for the summer

This will be a moving target, but it seems like after the first week these may be good goals for the summer:

  • A ( pkgdown) vignette for the grader user API and a separate one for the developer API.
  • Blog post of how learnr works, and what order to expect error messages (see use-case maps).
  • Put together a grader example without the use of learnr
    • An example would be to pass grader 2 R scripts (the student code and the solution code).
    • This came up while I was talking to Greg and he brought up “Working Effectively with Legacy Code” by Michael Feathers
      • This will help me find the entry points to the package, write isolated tests, and have a clearer goal of the project.
      • Along with use-case maps, this will help me be a better software developer and understand what’s actually going on (did I mention I’m half panicked and fighting imposter syndrome?)

Other goals would be to have:

  • Virtual coffee meetings with other people in the company (there’s a channel called “virtual-donut” that has a scheduler bot)
  • See if I can draft up and hunt down the people for a PhD topic.
    • Greg really planted a seed in my head a few months ago about putting together materials to help “front-line health practitioners (e.g., nurse and general practitioners) use and understand data”.
    • I think I can… I think I can… I think I can…

Random things I’ve learned so far

  • I think I really like this 100% remote work schedule, I’ve been up way earlier than usual finishing things before a morning meeting and then taking my time the rest of the day planning and implementing things.
    • I also feel like a teenager in chatrooms talking to everyone (thanks Slack!)
  • Using Templates with Roxygen2: allows roxygen2 docstrings to be reused by using a @template Rd keyword
  • Use-case maps: document and show the various components of an application, but also show what components are used for a given action
    • This would be great for learnr and grader since I’ll pretty much be doing this for myself just to figure out all the moving parts of the system and how each component works in the full application.
  • lintr: is a way to do static code analysis in R.
    • You can use this to check for (potential) errors before runtime
    • Pretty sure I can take the same idea and apply it to grader
  • All these years practicing my Git-fu seems to be paying off.
  • My mentor, Garrett Grolemund, is RStudio employee number insert number that can be represented with 1 hand :O
  • There’s an “emoji-psa” channel at RStudio. It’s active. People really know their emojis. owo

Misc

  • I’ve given up trying to sync the GSuite with Thunderbird and having a unified calendar.
    • I’m using openWMail again for my RStudio calendar/drive.
    • Still have all my emails in Thunderbird.
  • Not all the interns started on the same date. Most of us started June 3rd, but some started earlier and others will start later.
    • Us interns have our own slack workspace!
    • We had our first interns-only coffee chat on Friday.
      • Malcolm Barrett, Julia Blum, Joyce Cahoon, Desiree De Leon, Dewey Dunnington, and Maya Gans all met this week.
      • We’re planning to have an intern “host” one each week so everyone gets a chance to meet some (if not all) of us.
  • Allison Hill is hosing a “Summer of blogdown” this week, so I’ll probably be doing another website update.
Posted on:
June 10, 2019
Length:
5 minute read, 896 words
Tags:
R rstudio-internship
See Also:
R Hex Bowtie
My time as an RStudio Intern
Table of Model Results using kable and kableExtra