February 13, 2019 by @khondula
Putting your app code and any associated files (data, images, etc.) in a GitHub repository makes your app available for anyone with R installed to run it from their local R session by using the runGitHub function in the shiny package or downloading the repository. As with any GitHub repository, this also means you can also “release” a version of the app to get a citable DOI using Zenodo.
Whereas SESYNC’s shiny server is only designed for prototyping and testing, RStudio’s shinyapps.io is available as a longer term solution. There is a free tier that allows limited web traffic (up to 5 applications and 25 hours per month), and pay plans that offer more traffic, apps, and support. Hosting on other cloud services like AWS or Digital Ocean will require more configuration but can also be used to run your app.
Another possibility is sharing visualizations or reports as rendered HTML pages that don’t require R code to run. For example, interactive visualizations like the output from leaflet maps, dygraphs, timevis, or R Markdown can be saved as HTML files that can be opened up in a web browser like Firefox or Chrome. They can also be hosted on GitHub as websites based on the repository name by enabling the GitHub pages option in the repository Settings.
Your audience will have a more satisfying user experience if your app isn’t unnecessarily slow to load or display changes. If you are looking to make improvements, the first step is to figure out where the bottlenecks are. RStudio’s Shiny website has several in-depth articles on how to improve performance through plot caching and load testing, but a few initial steps to take are to: a) do as much filtering and summarizing data outside of shiny as possible, and b) save really large data files as feather files instead of csv to read them in faster.