If a value change in shiny - r

Question

I seek a method in R shiny that I can include inside a render or an observe to check if a certain value has changed.
For example :
library(shiny)
library(shinycssloaders)
ui <- fluidPage(
fluidRow(
column(5,
imageOutput("image") %>% withSpinner()
),
actionButton("button", "redo")
)
)
server <- function(input, output, session) {
data = reactiveVal(data.frame(
gp = factor(rep(letters[1:3], each = 10)),
y = rnorm(30)
))
getWidth = function(image)
paste0(session$clientData[[paste0("output_", image, "_width")]], "px")
output$image = renderImage({
input$button
outfile = tempfile(fileext = ".png")
p = ggplot(data(), aes(gp, y)) +
geom_point()
Sys.sleep(2) # to symbolise a plot which is very slow to appear
ggsave(filename = outfile, p)
return(list(src = outfile, width = getWidth("image")))
}, deleteFile = F)
}
shinyApp(ui, server)
Here just when I resize the window, the image is resaved, I do not want that. But I want that code to save the file if and only if data() or input$button is changed.
The only solution I see so far is to copy the data in an independent variable and to check if the value has changed. If the data change, save the new plot, change the value of the independent variable.
But I am not convinced that is it a good solution because the value data will be copied twice. For this dataset it not very severe, but for a dataset with millions lines the strain is harder. Or a graph that takes more than 10 seconds to save.
Thank you,

Answer 1

My suggestion would be
Use renderPlot instead of renderImage
Create the plot in a reactive expression
Save only when the plot changes (now it only reacts to data changes not to resizes) or the button is pressed, by using an observeEvent with those two events as triggers.
Find a working example below. If you want to change the size of the saved plot do it in the ggsave.
library(shiny)
library(shinycssloaders)
library(tidyverse)
ui <- fluidPage(
fluidRow(
column(5,
imageOutput("image") %>% withSpinner()
),
actionButton("button", "redo")
)
)
server <- function(input, output, session) {
data = reactiveVal(data.frame(
gp = factor(rep(letters[1:3], each = 10)),
y = rnorm(30)
))
p <- reactive({ggplot(data(), aes(gp, y)) +
geom_point()
})
observeEvent(c(p(), input$button), {
outfile = tempfile(fileext = ".png")
ggsave(filename = outfile, p())
})
output$image = renderPlot({
Sys.sleep(2) # to symbolise a plot which is very slow to appear
p()
})
}
shinyApp(ui, server)