I have a very long task in R that does a lot of computations and writes out several files. This function is called from within a Shiny-reactive. Since it takes so much time I wanted to parallelize it on several processes but at the same time I wanted to include a progress-bar for the end user. I closely followed the answer to this post that uses futures
: Using standard R shiny progress bar in parallel foreach calculations .
However: The files that my function creates are later used to do some further processing in another reactive. Somehow I am unable to let this second reactive wait until my future has fully completed - it already tries to read files that either are not yet present or incomplete. How do I tell the second reactive to wait until the future is complete?
My code so far:
library(promises)
library(future)
library(ipc)
plan(multisession)
r1 <- eventReactive(input$go_Button, {
Runs <- 1:3 #3 parallel runs
prog <- list() #Setup for Progressbar
for (j in Runs) {
prog[[j]] <- AsyncProgress$new(message = "Calculation")
future_promise({
for (k in 1:length(rawdata_entries)) {
prog[[j]]$inc(1 / length(rawdata_entries))
<<< VERY LONG TASK THAT CREATES SEVERAL FILES >>>
}
prog[[j]]$close()
}) # end of future_promise
} # end of for loop
return(list("Done"=TRUE))
}) # end of eventReactive
r2 <- reactive({
if (r1()$Done == TRUE) { #my naive approach that doesn't work
<<<do some further processing with the files r1() created >>>
<<< creates the data-frame "some_results" >>>
return(some_result)
}
})
The most important things I want to achieve are:
- parallelise the time consuming calculations
- have a progressbar for the end user
- fully evaluate the promise before moving on with further steps
Where is / are my mistake(s)?