Updating a reactive 'value' using itself in R Shiny

I am trying to build a dashboard to keep track of my groups bugs in bugzilla. The query to retrieve this data is slow, so I would like to only retrieve changed bugs and update a local copy.

I have a function 'get_bugzilla' which returns everything, or if provided a timestamp, everything changed after that timestamp.

My current best attempt to use this in a reactive fashion is:

poll<-reactiveTimer(intervalMs = 10000)
ckbdata<-reactive({get_bugzilla()})
ckbdata<-reactive({
	poll()
	wip<-ckbdata()
	new<-get_bugzilla(max(wip[['last_change_time']]))
	if(length(new)>0){
		wip<-wip[!(id %in% new[['id']]),]
		wip<-rbind(wip,new)
	}
	wip
})

This is producing the error "evaluation nested too deeply: infinite recursion / options(expressions=)?", which was something I was worried about. But I cannot work out the correct way of doing this.

This is because there is an infinite loop i.e. one of the reactive expression is calling another and vice versa. ckbdata is defined as a reactive expression twice, which is likely causing the infinite recursion.

Maybe this works

poll <- reactiveTimer(intervalMs = 10000)

ckbdata <- reactive()
  poll()
  current_data <- isolate(ckbdata())
  # retrieve all the data, if the current data is empty
  if (is.null(current_data)) {
    get_bugzilla()
  } else {
    get_bugzilla(max(current_data[['last_change_time']]))
  }
})
ckbdata_updated <- reactive({
  current_data <- ckbdata()
  # Update the local copy in case of modification or bug
  new_data <- get_bugzilla(max(current_data[['last_change_time']]))
  if (length(new_data) > 0) {
    current_data <- current_data[!(current_data$id %in% new_data[['id']]),]
    current_data <- rbind(current_data, new_data)
  }
  current_data
})

Cheers

The solution appears to be fairly easy, despite being a challenge to find good examples. The key thing was to store the results of the first slow request as part of the creation of the reactiveValues, and to have an isolated copy of that inside the observe so that it can check if there are any new rows, and process them accordingly.

poll<-reactiveTimer(intervalMs = 10000)

values<-reactiveValues(ckbdata=get_bugzilla())

observe({
	poll()
	wip<-isolate(values$ckbdata)
	new<-get_bugzilla(max(wip[['last_change_time']]))
	if(length(new)>0){
		wip<-wip[!(wip$id %in% new[['id']]),]
		values$ckbdata<-rbind(wip,new)
	}
})