shiny: manipulating global variables to input back into shiny - shiny

Question

I have made a shiny app the removes outlying experimental artefacts:
df=data.frame(seq(-1000,1000), rnorm(2001)) #not real data
ui <- basicPage(
plotOutput("plot1", click = "plot_click", brush = "plot_brush"),
verbatimTextOutput("info"),
plotOutput("tab")
)
server <- function(input, output) {
output$plot1 <- renderPlot({
plot(df[,1], df[,2])
})
data_new<-eventReactive(input$plot_brush,{
da=df
rowmin <- which(da[,1] == round(as.numeric(input$plot_brush$xmin)))
rowmax <- which(da[,1] == round(as.numeric(input$plot_brush$xmax)))
da[rowmin:rowmax,2] = mean(da[,2])
da=isolate(da)
#writeToDatabase(da)
})
#output$zoom<-renderable({plot(data_new()[,1], data_new()[,2])})
output$tab=renderPlot({plot(data_new()[,1], data_new()[,2])})
}
shinyApp(ui, server)
This works fine but it is inconvenient that I can not permanently remove the artefacts i.e. I was wondering is there any way to make the values of the non-reactive variable permanently retain the changes made rather than the original incorrect data frame being replotted each time?
I have tried using a function that corrected the faulty data-variable 'df':
change=reactive(function(d){
rowmin <- which(d[,1] == round(as.numeric(input$plot_brush$xmin)))
rowmax <- which(d[,1] == round(as.numeric(input$plot_brush$xmax)))
d[rowmin:rowmax,2] = mean(d[,2])
return(isolate(d))
})
isolate(change(df))
but I get the following error:
Listening on http://127.0.0.1:6183
Warning: Error in change: unused argument (df)
Stack trace (innermost first):
52: change
51: is.data.frame
50: write.table
43: isolate
42: server [/Users/Downloads/test.R#20]
1: runApp
Error in change(df) : unused argument (df)
This was a starter test to see if I could dynamically update the variable. All of this is with the aim of being able to successively view and eliminate each of the sharp error peaks in my data shown above- rather than the code re-running on the same immutable (from shiny's perspective) variable each time??

Answer 1

You probably want to use reactiveValues:
server <- function(input, output) {
my <- reactiveValues(df=data.frame(seq(-1000,1000), rnorm(2001))) # Initialize df
output$plot1 <- renderPlot({plot(my$df[,1], my$df[,2])})
observeEvent(input$plot_brush,{
rowmin <- which(my$df[,1] == round(as.numeric(input$plot_brush$xmin)))
rowmax <- which(my$df[,1] == round(as.numeric(input$plot_brush$xmax)))
my$df[rowmin:rowmax,2] <- mean(my$df[,2]) # Update df
})
}