We’re excited to announce the release of Shiny v1.3.2. This release has two main features: a new reactivity debugging tool we call
reactlog, and much faster serving of static file assets.
Debugging faulty reactive logic can be challenging, as we’ve written and talked about in the past. In particular, some of the most difficult Shiny app bugs to track down are when reactive expressions and observers re-execute either too often (i.e. plots that render multiple times in succession after a single change), or not often enough (i.e. outputs that don’t update when you expected them to).
This release has an important new addition to the Shiny debugging toolbox:
reactlog! To use
reactlog, execute this line before running your Shiny app:
options(shiny.reactlog = TRUE)
This will instruct Shiny to keep a record of all the interactions between reactive objects.
Then, use your app, reproducing the problematic symptoms. Once you have done that, press
Ctrl+F3 (Mac users:
Cmd+F3) from within your browser, and you’ll see something like this:
This screen lets you interactively explore the reactive history of your Shiny session. You can step forwards and backwards through time, watching as reactive objects execute, create and sever relationships, invalidate, etc.
For medium and large Shiny apps, the reactive graph may be pretty crowded when visualized in two dimensions. Two
reactlog features help you separate the signal from the noise.
Together, these features make it easy to find and focus on the relevant objects in your app.
With the new version of Shiny, static files are always served up at lightning speed, no matter what’s going on in R. We accomplished this by adding new static-file serving options to
httpuv, using dedicated C++ code paths running on a background thread. This means that computations in R won’t affect the serving of static files, and serving static files won’t affect computations in R. The experience for users of heavily-loaded Shiny applications should be noticeably better. Note that it has always been possible with RStudio Connect and Shiny Server Pro to improve performance by increasing the number of R processes serving an application, but now Shiny itself is more efficient and multithreaded, so each R process can effectively handle more user sessions.
The best part is that you don’t need to do anything to take advantage of these speed improvements—just upgrading Shiny to v1.3.2 will do it!
Note: A number of users have reported that upgrading to Shiny v1.3.0 (or higher) breaks their apps when running behind an Nginx proxy: the HTML loads, but none of the styles are applied and none of the calculations run. This occurs when Nginx is subtly misconfigured. We’ve posted details and a fix in this RStudio Community post.
Welcome to the rstudio::glimpse() newsletter. Get a glimpse into our tools and how to use them.
We’d like to highlight and answer some of the great audience questions asked during the "MLOps with vetiver in Python and R" webinar.