Are the two chunks of code
logTimer <- reactiveFileReader(intervalMillis = 500, session=NULL,
filePath = logFile, readFunc = readLines)
output$logf <- renderPrint({
logTimer()
})
and
output$logf <- renderPrint({
reactiveFileReader(intervalMillis = 500, session=NULL,
filePath = logFile, readFunc = readLines)()
})
Equivalent in shiny ?
they are not perfectly equivalent if only for the reason that one creates an object named logTimer and the other doesnt.
Of course, but I wonder if creating the reactiveFileReader directly inside renderPrint is erroneous. Here is a (somewhat lengthy) chunk of code where it does not work whereas using a reactive variable does.
library(shiny)
library(processx)
asyncUI <- function(id) {
ns <- NS(id)
tagList(
fluidRow(
actionButton(ns("runProc"), "Sleep"),
verbatimTextOutput(ns("logf")),
),
)
}
asyncServer <- function(id, logFile)
moduleServer(id, function(input, output, session) {
mainProcess <<- NULL
observeEvent(input$runProc, {
mainProcess <<- process$new("count.tcsh")
})
procTimer <- reactivePoll(1000, NULL,
checkFunc = function() {
if(is.null(mainProcess))
NULL
else
mainProcess$is_alive()
},
valueFunc = function() {
if(is.null(mainProcess))
NULL
else
mainProcess$is_alive()
})
# logTimer <- reactiveFileReader(intervalMillis = 500, session=NULL, filePath = logFile, readFunc = readLines)
output$logf <- renderPrint({
req(!is.null(procTimer()))
reactiveFileReader(intervalMillis = 500, session=NULL, filePath = logFile, readFunc = readLines)()
#logTimer()
})
})
ui <- fluidPage(
asyncUI("myrun"),
)
server <- function(input, output, session) {
asyncServer("myrun", logFile="count.log")
}
shinyApp(ui, server)
This runs a script count.tcsh that writes to a file counts.log. It displays the log in real time. count.tcsh script is
#!/bin/tcsh
echo "Begin script" >! count.log
foreach i (`seq 1 10`)
sleep 1
echo $i seconds passed >> count.log
end
echo Finished script >> count.log
When I run it (in Linux), the version above does not display all the log file counts from 1 to 10, but (for a reason unknown to me) stops at count 2-3. If I use a variable inside renderPrint, it works just fine.
Myself I would prefer to be practical and write the cleanest easiest to follow code that I know how to do.
Though I do completely get the desire to understand the framework, and I have to admit I don't have the chops to do that for you ...
Of the different options I show below. 1 and 3 makes sense to me, and I have used them often in the past, and 2 seems to follow your imagined pattern, but I wouldn't have thought to use it.. maybe a lack of imagination on my part.. , and I don't have a clear reason for you as to why it doesn't work, you may need someone with knowledge of shiny internals to figure it for you, I expect there's something about it that does not reliably build the reactive graph.
Anyways, you have a solution so its not a big defeat, simply revert to using your logTimer object.
Example of problematic pattern:
library(shiny)
ui <- fluidPage(
actionButton("mybtn","press"),
verbatimTextOutput("mytext1"),
verbatimTextOutput("mytext2"),
verbatimTextOutput("mytext3")
)
server <- function(input, output, session) {
some_named_reactive <- reactive({input$mybtn})
output$mytext1 <- renderPrint({
some_named_reactive()
})
output$mytext2 <- renderPrint({
# best avoided
reactive({input$mybtn})()
})
output$mytext3 <- renderPrint({
input$mybtn
})
}
shinyApp(ui, server)Thank you for the thinking.
By the way, your reprex is much clearer than my long example.