Ungroup & Select in shiny unexpected behaviour

Question

I have a shiny app where the ungrouping is not working properly when I compare the result from "un-shiny" version of the output with the tableOutput in Shiny. I was expecting both to have a same result. Please see this for reference:

# "un-shiny" version to compare results, not part of the app. 
# This is to match with mpg input in shiny
mtcars %>%
  select(mpg, cyl, disp,
                wt, gear) %>%
  mutate(wt = as.integer(wt),
         mpg =as.integer(mpg)) %>% 
  filter(gear >= 4 & gear <= 5) %>%
  group_by(wt,gear, mpg) %>%
  tally() %>%
  ungroup() %>%
  group_by(gear,mpg) %>%
  rename(Count= n)%>%
  mutate(summ = sum(Count),
         Percentage = round((Count/summ)*100, 2))

This gives me this output-

Now, in the shiny app-

library(dplyr)
library(tidyr)
library(shiny)
  

#global.R -----


mtcars_rdpr <- mtcars %>%
  select(mpg, cyl, disp,
                wt, gear) %>%
  mutate(wt = as.integer(wt),
                mpg =as.integer(mpg)) %>% 
  filter(gear >= 4 & gear <= 5) %>%
  

#ui.R ---- 
ui <- fluidPage(
  sidebarLayout(
    mainPanel(
      
      selectInput("pickvalue", label = "Gears", colnames(mtcars_rdpr%>%
                                                          select(mpg, cyl)),
                  selected = NULL, multiple = F)),
    
    
    tableOutput("tableOut")
    
  )
)

# server.R-----
server <- function(input, output, session){
  
  
  gears <- reactive({
    
    dat <- mtcars_rdpr
    
    if (!is.null(input$pickvalue)){
      dat <- dat %>%
        group_by(wt,gear, input$pickvalue) %>%
        tally() %>%
        ungroup() %>% # when i mention specific columns in the function, I get a warning
        group_by(gear,input$pickvalue) %>%
        rename(Count= n)%>%
        mutate(summ = sum(Count),
               Percentage = round((Count/summ)*100, 2))
    } 
    
    #dat <- dat %>% select(-input$pickvalue) # this gives a warning
    
    return(dat)
    
  })
  
  output$tableOut<- renderTable({gears()})
  
}

shinyApp(ui = ui, server=server)

I would really appreciate it if someone can explain to me why the results are not matching between the two.

Answer 1

You cannot mix up non-standard evaluation with character strings, which you essentially did in your code.

As of dplyr 1.0.0, you can use across and combine them in a vector:

Just the relevant line:

  group_by(across(c("gear", input$value))) %>%