Strange actual parameter assignment in shiny's renderPlot - shiny

Question

When using a for loop to render some plots I obtained results I didn't understand. All plots became equal to the (intended) last one. I hope to make this more clear in the example below. Note the plot1 shows a '6' (the last one), not a '5' (as intended).
Can anyone explain this and/or give a solution avoid this kind of behaviour.
shiny ui:
shinyUI (fluidPage(
fluidRow (
column (6, plotOutput ('plot1', height = "180px")),
column (6, plotOutput ('plot2', height = "180px"))
),
fluidRow (
column (6, plotOutput ('plot3', height = "180px")),
column (6, plotOutput ('plot4', height = "180px"))
)
))
shiny server:
shinyServer (function (input, output, session) {
require (ggplot2)
plotId <- c('plot1', 'plot2', 'plot3', 'plot4')
pw <- initPlotWindows ()
output[['plot1']] <- renderPlot ({pw[[1]]})
output[['plot2']] <- renderPlot ({pw[[2]]})
output[['plot3']] <- renderPlot ({pw[[3]]})
output[['plot4']] <- renderPlot ({pw[[4]]})
id <- 'plot1'
i <- 5
output[[id]] <- renderPlot ({pw[[i]]})
id <- 'plot3'
i <- 6
# output[[id]] <- renderPlot ({pw[[i]]})
})
initPlotWindows <- function () {
pw <- vector (mode = 'list', length = 6) # plot window data
for (i in 1:6) pw[[i]] <- p_empty (plotN = i)
pw
}
p_empty <- function (plotN) {
p <- ggplot()
p <- p + annotate("text", label = plotN, x = .5, y = .5, size = 20, colour = "grey")
p <- p + theme(axis.ticks.x = element_blank(), axis.text.x = element_blank(), axis.title.x=element_blank())
p <- p + theme(axis.ticks.y = element_blank(), axis.text.y = element_blank(), axis.title.y=element_blank())
p
}
Resulting output. Plot1 contains a 6 not a 5!

Answer 1

The reason why this happens is because shiny does not execute your server side code line by line. Shiny executes all of the non-reactive code once at the beginning. But renderPlot() and other reactive code will only run when it is necessary. (when they are called)
So what happened is, it assigned 5, and then 6 to i, and only rendered the plot afterwards.
output[[id]] <- renderPlot ({
print("section D")
pw[[i]]})
You can check it pretty easily in the R console using the print() function:
shinyServer (function (input, output, session) {
print("section A")
require (ggplot2)
plotId <- c('plot1', 'plot2', 'plot3', 'plot4')
pw <- initPlotWindows ()
output[['plot1']] <- renderPlot ({
print("section B1")
pw[[1]]})
output[['plot2']] <- renderPlot ({
print("section B2")
pw[[2]]})
output[['plot3']] <- renderPlot ({
print("section B3")
pw[[3]]})
output[['plot4']] <- renderPlot ({
print("section B4")
pw[[4]]})
print("section C")
id <- 'plot1'
i <- 5
output[[id]] <- renderPlot ({
print("section D")
pw[[i]]})
print("section E")
id <- 'plot3'
i <- 6
# output[[id]] <- renderPlot ({pw[[i]]})
})
Console output:
[1] "section A"
[1] "section C"
[1] "section E"
[1] "section B1"
[1] "section B2"
[1] "section B3"
[1] "section B4"
[1] "section D"
Edit:
As for the output[[id]], id gets evaluated at the beginning, so output[[id]] becomes output[[plot1]]. But pw[[i]] is only evaluated when shiny asks the plot to render, which is after the value has been set to 5 and then to 6.
If you want plot1 to show 5, then you could assign id outside of renderPlot(), and assign i inside of renderPlot():
id <- 'plot1'
output[[id]] <- renderPlot ({
i <- 5
pw[[i]]})

Answer 2

Thank you GyD,
From your answer I conclude that a normal for-loop is not possible. The running variable 'i' would than be defined outside and not inside renderPlot().
So it seems that I have to use eval (parse()) for at least renderPlot().
A for-loop like:
for ( i in 1:4) {
pltId <- paste ('plot', i, sep='')
output[[pltId]] <- renderPlot ({pw[i]})
}
would then become (using eval(parse()) for the whole assignment):
for ( i in 1:4) {
eval (parse (text = paste ("output[['plot", i, "']] <- renderPlot ({pw[", i, "]})", sep='')))
}

Answer 3

Your answer should work, but you could generate both the UI elements and the server side with loops. (don't know if that interests you)
I am not a huge fan of for loops in R, so I'm using lapply.
library(shiny)
library(ggplot2)
# I want to plot these elements
vec <- c(5, 2, 3, 4)
# Defining some functions
initPlotWindows <- function () {
pw <- vector (mode = 'list', length = 6) # plot window data
for (i in vec) pw[[i]] <- p_empty (plotN = i)
pw
}
p_empty <- function (plotN) {
p <- ggplot()
p <- p + annotate("text", label = plotN, x = .5, y = .5, size = 20, colour = "grey")
p <- p + theme(axis.ticks.x = element_blank(), axis.text.x = element_blank(), axis.title.x=element_blank())
p <- p + theme(axis.ticks.y = element_blank(), axis.text.y = element_blank(), axis.title.y=element_blank())
p
}
# Creating pw list
pw <- initPlotWindows()
ui <- fluidPage(
# Creating UI from loop
lapply(vec, function(i) {
column(6, plotOutput(paste0('plot', i), height = "180px"))
})
)
server <- function(input, output, session){
# Generating UI output from loop
lapply(vec, function(i) {
output[[paste0('plot', i)]] <- renderPlot({
pw[[i]]
})
})
}
shinyApp(ui, server)