Insert row get slower in shiny WITH an observeEvent on the input$[name of the table]

[dominiqueemmanuel]

Hi,

Thanks for this R package !!

When I use an rhandsontable in a shiny application with an event observer, if I insert rows in the table ([right click] on a row then [insert row below]) the time spent to insert the row get slower and slower (after 10 insertions, the time is long, and avec 20 insertions the time is longer, etc...).

NB : if I remove the observeEvent lines, this is not happening anymore.

Do you have an idea how to solve this problem ?

Best regards,
Dom

Here is an example :

library(shiny)
library(rhandsontable)
ui<-fluidPage(
  rHandsontableOutput("aa")
)

server = function(input, output) {
  verbatim_rv<-reactiveValues()
  n<-100
  verbatim_rv$xx<-data.frame(c1="a"%>%paste0(seq(n)),c2="a"%>%paste0(seq(n)),c3="a"%>%paste0(seq(n)),c4="a"%>%paste0(seq(n)),c5="a"%>%paste0(seq(n)))

  ## When removing these lines, the problems doesn't happen anymore
  observeEvent(input$aa,{
    verbatim_rv$xx <<- hot_to_r(input$aa)
  })

  output$aa <- renderRHandsontable({
    choices<-letters[1:10]
    rule_table<-verbatim_rv$xx 
    rhandsontable(rule_table) %>% hot_context_menu() %>%
      hot_col(col = c("c1"), type = "autocomplete",source = choices,strict = FALSE) %>%
      hot_col(col = c("c2"), type = "autocomplete",source = choices,strict = FALSE) %>%
      hot_col(col = c("c3"), type = "autocomplete",source = choices,strict = FALSE) %>%
      hot_col(col = c("c4"), type = "autocomplete",source = choices,strict = FALSE) %>%
      hot_col(col = c("c5"), type = "autocomplete",source = choices,strict = FALSE) %>%
      hot_table(contextMenu=TRUE)
  })
}
app<-shinyApp(ui, server)
runApp(app,launch.browser=TRUE)

Here is my session informations :

> sessionInfo()
R version 3.3.1 (2016-06-21)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

locale:
[1] LC_COLLATE=French_France.1252  LC_CTYPE=French_France.1252   
[3] LC_MONETARY=French_France.1252 LC_NUMERIC=C                  
[5] LC_TIME=French_France.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] rhandsontable_0.3.2 shiny_0.13.2       

loaded via a namespace (and not attached):
 [1] htmlwidgets_0.7 magrittr_1.5    R6_2.1.2        tools_3.3.1    
 [5] htmltools_0.3.5 yaml_2.1.13     Rcpp_0.12.6     jsonlite_1.0   
 [9] digest_0.6.10   xtable_1.8-2    httpuv_1.3.3    mime_0.5     

[jrowen]

I haven't heard of this issue before. If you can debug further and provide more detail around which method is causing the issue, I can take a quick look.

[dominiqueemmanuel]

Yes, it seems to be a mysterious issue.

I'm pretty sure the issue is caused by the incrementation of the reactive value (i.e. in my example observeEvent(input$aa, { verbatim_rv$xx <<- hot_to_r(input$aa) })).

In practice I can manage to reduce the gravity of the issue by preventing unnecessary incrementations, but the problem is still happening...

[dominiqueemmanuel]

For information, I have tried to trace each part of the code (with a system.timecall and print) : every call of the code seems not to take more time while the number of inserted rows increases.

However, the increasing time seems to come from javascript...

[jrowen]

Does the issue appear to like with the rhandsontable, handsontable or
htlmwidgets functionality?

On Wed, Aug 10, 2016 at 8:37 AM dominiqueemmanuel notifications@github.com
wrote:

For information, I have tried to trace each part of the code (with a
system.timecall and print) : every call of the code seems not to spend
more time while the number of inserted rows increases.

However, the increasing time seems come from javascript...

—
You are receiving this because you commented.

Reply to this email directly, view it on GitHub
#98 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ABdYrtP5eyXKfat0aolvXVVlswvhDxk-ks5qedQDgaJpZM4JgUpm
.

[dominiqueemmanuel]

I don't think the issue comes (directly) from handsontable, because only inserting rows does not cause the issue (the issue only begins to appears if I add the line observeEvent(input$aa, { verbatim_rv$xx <<- hot_to_r(input$aa) }) ).

I don't know exactly where the time is used.

However, if I trace with options(shiny.trace = TRUE), I can see that the time used is between
SEND {"errors":[],"values":{"aa":{"x":{"data":[{"c1":"a1","c2":"a1","c3":"a1","c4":........
and
RECV {"method":"update","data":{"aa":{"data":[["a1","a1",......

[govertbuijs]

We're running into a similar problem using 0.3.3.
To debug : add the following code to rhandsontable.js, starting at line 84 (top of x.afterChange):

if (x.foo === undefined) {
  x.foo = new Date();
}
console.log(x.foo);

The list increases on every edit when there's an eventObserver on the table.
Somehow additional data objects keep getting created in JS, but i can't find who holds them.

[jrowen]

Thanks for filing an issue. This looks to be an issue with handsontable (or our implementation thereof).