Update data in shiny application without refreshing the whole application - r

Question

I have an application for real time data visualization build with R shiny library. I do periodic data reload from file using reactivePoll function. What I do not like about this is that whenever data reloads the whole application refreshes.
So for example if I have DT table output with selection and I use this selection input$table_rows_selected it resets to NULL whenever data reloads which is not user-friendly at all.
Is it overall possible to change data output without interrupting user?
UPDATE.
Can this be achieved with any other package for displaying tables - googleVis or other?
Working example.
library(shiny)
library(DT)
runApp(shinyApp(
ui = fluidPage(dataTableOutput('table')),
server = function(input, output, session) {
pollData <- reactivePoll(4000, session,
checkFunc = function(){ Sys.time() },
valueFunc = function(){ data.frame(id = sample(letters[1:3]), a = runif(3), b = runif(3), c = runif(3)) })
output$table <- renderDataTable({pollData()})
proxy <- dataTableProxy('table')
observeEvent(pollData(), {
selectRows(proxy, input$table_rows_selected)
})}
))
I have taken this example from #NicE answer and added id column. The point is that #NicE answer is OK if one needs certain row to be selected when that row is identified by the row number.
Now suppose I need a row to be selected when that row is identified by some id value. That is if I select a row with id equal b, then the next time data reloads I want the row to be selected with the same id value.

Answer 1

You could use a dataTableProxy to select rows when the datable is created after a pollData update.
Here's an example, the dataframe is refreshed every 4 seconds:
library(shiny)
library(DT)
ui <- fluidPage(dataTableOutput("table"))
server <- function(input,output,session){
values <- reactiveValues()
pollData <- reactivePoll(4000, session,
checkFunc=function(){
Sys.time()
},
valueFunc=function(){
data.frame(a=sample(c("a","b","c"),3),b=runif(3),c=runif(3),stringsAsFactors = F)
})
output$table <- renderDataTable({ pollData()})
observe({
values$selected <- pollData()$a[input$table_rows_selected]
})
proxy = dataTableProxy('table')
observeEvent(pollData(),{
selectRows(proxy, which(pollData()$a %in% values$selected))
})
}
shinyApp(ui,server)
Update: on the above code, when the data changes, the selected rows are the ones that have the same first column as previously.