VSCode vs RStudio — Worth the Switch?
Does Microsoft’s flagship code editor hold up against the old favorite?
RStudio is the preferred IDE/Editor for running R code for most people, but VSCode has rapidly become one of the most popular code editors for a huge number of other languages. This writeup is for people who are very familiar with RStudio and want to see how the two compare. So, I will not be talking a lot about RStudio for that reason, but going into the strengths and weaknesses of VSCode compared to RStudio.
Further, this is a very brief overview comparing the community (free) versions of both editors, I cannot speak to the R-Studio paid editions.
I am somewhat of a intermediate programmer who doesn’t use R as much as some people but whenever I do, RStudio is my preferred editor of choice. For everything else (Python, Jupyter, HTML, CSS, JS,etc ), I tend to use VSCode. I decided to try VSCode for R because I wanted to see if I could use a single code editor for everything I do.
RStudio is an outstanding piece of software and has never given me any major problems during my workflows, this was just an experiment to find a one-size-fits all code editor/IDE.
If you are only interested in my conclusion or the TLDR; :
No, I will not be switching to VSCode for R — FOR NOW.
But I love how great R looks and works on it already.
Now that we’re done with that, lets get into some of the details:
Installation
Assuming you already have R installed on your system, Installing VSCode and R is very straightforward;
- download the .pkg or .exe file and install the code editor;
- install the `languageserver` package for R (install.packages(“languageserver”)
- install the R extension from the marketplace (https://marketplace.visualstudio.com/items?itemName=REditorSupport.r R — Visual Studio Marketplace)
and you’re ready to start coding.
Optional: You can install a better terminal alternative like radian, a debugger and a better plot-viewer like httpgd because the builtin plot-viewer for VSCode isn’t a particularly good one.
Setting up the editor
You can now start working with code but there are potentially times you could run into errors while running your code. To eliminate any potential errors/code not work, you must make sure the R path is specified in the VSCode preferences (cmd+shift+p on mac) and search for R path.
Finding the path is simple, open R and type:
R.home(“bin”) and copy the output path.
Another somewhat major tweak you would want to do to the editor:
Some simple shortcuts like the pipe operator doesn’t work out of the box, so adding a shortcut for “%>%” or “|>” depending on your preference can be done using the following method:
Working with R code:
Here is a comparison of my two setups side by side with the same file and environment
RStudio:
R in VSCode:
Working with R is very similar to a regular RStudio IDE; R-extension provides the same support for installing packages, plots and plot viewer, global environment, datasets, plots, lists, variables, checking loaded packages etc.
Terminal and console position can be modified to make get the input and output panes next to each other, something you cannot do well in R-Studio. So VSCode feels very familiar and running code is highly intuitive.
Running Code
Running a piece of code is again, very similar to RStudio. A run file button is located at the top of the panel and individual lines of code can be run using the cmd+return shortcut on mac. The shortcut can be easily modified similar to how the pipe operator shortcut is modified above.
In terms of running and getting results for plain R code and .R files; VSCode is a great alternative for a lot of people including me. So if you’re like me and you use R mainly for statistical analysis, data visualization and data analysis; give VSCode a try.
Working with Notebooks and .Rmd
This is where VSCode falls so much behind RStudio it becomes a one horse race. The notebook support in VSCode is in a single word — terrible. Running code chunks in a .Rmd notebook feels and looks dated. My existing notebooks created in RStudio were buggy in VSCode despite not showing any errors in RStudio. Creating new notebooks is unintuitive and still a long way behind RStudio in terms of the overall feel to it.
I have searched for solutions, including globally installing pandoc, knitr and rmarkdown etc and trying multiple versions of R and the packages mentioned previously to no avail. This is one of the biggest reasons I am not completely switching to VSCode just yet.
A Note About Quarto
I recently started playing around with Quarto, which looks like the natural successor to Rmarkdown. So far, I have only used Quarto in RStudio and am a big fan of the way it looks and works. I have yet to try it on VSCode, so I cannot comment on it.
What I love about R in VSCode
VSCode is faster, quick to load up, fast and snappy when the code is running. I did not do any analysis on the times it took to load a piece of code on one vs the other but VSCode ‘feels’ quicker. VSCode also functions as a one stop shop for R, Python, HTML, CSS, and dozens other languages and does it much better than RStudio. One of the most crucial features for a lot of R users is MULTIPLE R SESSIONS!!! — which comes with VSCode , a free software. It’s especially helpful when you’re trying to develop several related projects at the same time. Code Refactoring in VSCode is also much better than RStudio
Color picker and color blocks are highly useful for people like me who work with a lot of data visualizations, the ability to see the colors and use the color blocks to pick colors seamlessly is a game-changer.
Language server and document outline features help in navigating through a long line of code and make working with multiple R files painless. Intellisense and code completion is a lot faster in VSCode and more user friendly.
R-documentation and help is handled very differently in both the softwares. When it comes to help in VSCode, the information is available much more quickly and the information is a lot more streamlined.
Similarly , working with teams and multiple people on a single R file is easier using Microsoft’s Liveshare. Git and github integration in VSCode is easily one of the best in any code editor out there, further helping when working with multiple people. Code Snippets are helpful when working with a function or line of code frequently. I personally found VSCode snippets easier to use and set up than RStudio. Functions are handled brilliantly. Hover over them and you get all kinds of information and callbacks. VSCode handles code organization brilliantly; code sections allow for folding of code, makes for easier code reading and navigation and cleaner code overall.
Lastly, the entire code editor is highly customizable and you can make it your own by editing the json files inside VSCode. As such; VSCode has some amazing features that could really tempt an RStudio user.
What I don’t love about R in VSCode
One of the biggest reasons any R-code uses R-studio is because it works great out of the box, and we do not need to install extensions while VSCode setup for R can be tricky and can sometimes involve trail and error. Furthermore; R-templates inside R studio are amazing, and easy to find and use. Data-viewer in RStudio also feels a lot better than VScode, especially when working with a large dataset.
Even though I use VSCode for Python, HTML, CSS etc, coding R in it, especially for the first few hours feels very odd and in a way unfamiliar despite the very familiar overall IDE organization. Currently the biggest reason not to switch to VSCode is NOTEBOOK SUPPORT — If you work with rmd more than r, DO NOT SWITCH; running chunks of code is very clunky and poor inside VSCode and often can lead to unproductive sessions of code. Additionally, knitting notebooks doesn’t work well at all, and can lead to errors a lot of the time.
Even with the httpgd` package, plot viewers for VSCode are not on par with the in-built plot viewer of RStudio. R-terminal will sometimes throw errors in VSCode that you wouldn’t find for the same piece of code in RStudio and something that you might not want to deal with when you’re in a hurry.
In essence, RStudio just feels a lot more tailored to a data scientist/statistical analysis person, while VSCode can feel like a developer-centric code editor.
Things I hope to try soon with VScode
- Quarto
- Rblogdown and Rbookdown
- More Debugging
- Radian terminal
Conclusion
Do I love R in VSCode so far? Absolutely
Enough to make me want to switch? Not yet.
Why? Notebooks
Will I switch soon? Hopefully yes.
If you liked this, and have any suggestions for me; find me on twitter:
Further Reading
- A great video by Kun Ren on the subject : https://youtu.be/9xXBDU2z_8Y
- A blog on the same subject: https://renkun.me/2019/12/11/writing-r-in-vscode-a-fresh-start/
- Running R in VSCode: https://www.infoworld.com/article/3625488/how-to-run-r-in-visual-studio-code.html
- Insalling R in VSCode: https://code.visualstudio.com/docs/languages/r
- Bonus: Quarto vs RMD: https://yihui.org/en/2022/04/quarto-r-markdown/
Thank you for reading. Happy coding!
