Workflow

Lecture 23

Dr. Eric Friedlander

College of Idaho
CSCI 2025 - Winter 2026

Development Workflow

Creating and Running

• Creation:
  • One file app: app.R
  • Two file app: ui.R and server.R
• In RStudio/Positron:
  • Click the Run App button (play icon) in the toolbar.
• Stopping:
  • Click the Stop sign in the console.

Shiny Skeleton

library(shiny)
ui <- fluidPage(
)
server <- function(input, output, session) {
}
shinyApp(ui, server)

Debugging

Three Types of Problems

1. Unexpected Error: You get a traceback.
   • Easiest to fix.
2. Incorrect Value: No error, but wrong result.
   • Requires investigation (browser()).
3. Incorrect Logic: Updates happen at wrong time (or not at all).
   • Hardest. Unique to Shiny’s reactivity.

Tracebacks and browser()

• Traceback: Read from bottom to top to find your code.
• browser():
  • Place browser() inside your server function.
  • Execution pauses when it hits that line.
  • You can inspect input, output, and intermediate variables in the console.

server <- function(input, output, session) {
  observeEvent(input$go, {
    browser() # Execution pauses here
    # Inspect input$x, calculated values, etc.
  })
}

Print Debugging

• Classic technique: Print values to see what’s happening.
• Use message() instead of print().
  • Goes to standard error (logs).
• Use glue::glue() for easy formatting.

library(glue)
server <- function(input, output, session) {
  val <- reactive({
    message(glue("Updating val. input$x is {input$x}"))
    input$x * 2
  })
}

Debugging Reactivity

• Reactivity is lazy and non-linear.
• Code doesn’t run top-to-bottom.
• Visualizing the Reactive Graph helps.
• Use message() inside reactive() to verify:
  1. Is it running?
  2. Is it running too often?
  3. Is it running in the expected order?

Getting Help

The Reprex

• Reprex: Reproducible Example.
• Critical for getting help (StackOverflow, RStudio Community).
• Properties:
  • Minimal: Remove everything not related to the bug.
  • Self-contained: Copy/pasteable. runs on anyone’s machine.
    • No read.csv("C:/MyFiles/data.csv").
    • Use code to generate dummy data.

Creating a Shiny Reprex

1. Isolate the problem (remove unrelated UI/Server code).
2. Use built-in datasets (e.g., mtcars).
3. Combine UI and Server in one file.

library(shiny)

ui <- fluidPage(
  numericInput("n", "N", 5),
  textOutput("txt")
)

server <- function(input, output, session) {
  output$txt <- renderText(input$n * 2)
}

shinyApp(ui, server)

Practice!

Let’s work through the case study in the text!

Summary

Workflow Tips

• Read tracebacks carefully.
• Use browser() to stop and inspect.
• Use message() to track reactive execution.
• When stuck, create a Reprex.

Do Next

1. Read Chapter 5: Workflow from Mastering Shiny.
2. There’s NO recitation Gem for this textbook but I recommend creating your own and adding the textbook chapter and these slides.
3. I reommend working through the Exercises in the textbook.
4. Move on to Lecture 23!