Cannot ref shiny reactiveValues about lobstr

[jcheng5]

I'm missing some context here but if you do decide to set the shiny.suppressMissingContextError option, please make sure to restore its previous value when you're done.

[hadley]

Can you please provide a minimal reprex (reproducible example)? The goal of a reprex is to make it as easy as possible for me to recreate your problem so that I can fix it: please help me help you!

If you've never heard of a reprex before, start by reading "What is a reprex", and follow the advice further down the page. Please make sure your reprex is created with the reprex package as it gives nicely formatted output and avoids a number of common pitfalls.

[dipterix]

Can't use reprex for the case 1 because it runs forever.

Case 1:

# devtools::install_github('beauchamplab/rave@rave-fir')
  library(rave)
  library(lobstr)
  # rave::download_subject_data('https://s3-us-west-2.amazonaws.com/rave-demo-subject/sfn-demo/data-large.zip')
  m = unlist(load_modules())[[3]]
  init_app(m, test.mode = T)
  ref(m)

Error message:

'as.list.reactivevalues' is deprecated. Use reactiveValuesToList instead.
Please see ?reactiveValuesToList for more information.
Error in .getReactiveEnvironment()$currentContext() : 
  Operation not allowed without an active reactive context. (You tried to do something that can only be done from inside a reactive expression or observer.)

Second case doesn't go well with reprex as well since the shiny app didn't launch.

suppressPackageStartupMessages(library(shiny))
env = new.env()

shinyApp(fluidPage(actionButton('asd',' ')), server = function(input,output,session){
  local_data = reactiveValues(a=1)
  
  env$sess = session
  stopApp()
})

Shiny applications not supported in static R Markdown documents

lobstr::ref(env$sess)
#> [1:0x7fada201e2e0] <NULL>

^{Created on 2018-12-05 by the reprex package (v0.2.1)}

If I ran without reprex, then the error message is:

>   lobstr::ref(env$sess)
Error in is.character(lhs) : 
  C stack usage  7970736 is too close to the limit
Error during wrapup: C stack usage  7969856 is too close to the limit

[hadley]

Can you please give me some snippet off code to run that doesn't require installing other packages or installing large data? It should be possible to make a very small shiny app to demonstrate the problem.

[yutannihilation]

I can reproduce the issue with the code above, though the error message is different now.

> lobstr::ref(env$sess)
Error in is.character(lhs) : node stack overflow

This part might be done with shiny:::ShinySession$new(ws), but I'm yet to find what ws I should pass...

  env$sess = session

[hadley]

@jcheng5 should be able to help you figure it out

[yutannihilation]

I can reproduce the issue with the code above

Sorry, I just looked at the second issue... 🙄

For the first case, @jcheng5 suggested (thanks!) we can use options(shiny.suppressMissingContextError=TRUE) to suppress the error.

For the second case, it seems there's a ciruclation of reference:

> unclass(env$sess$.__enclos_env__)
<environment: 0x00000197c79482c0>
> unclass(env$sess$.__enclos_env__$self$.__enclos_env__)
<environment: 0x00000197c79482c0>

[yutannihilation]

For the second case, it seems there's a ciruclation of reference

Sorry, this is not the core of the problem, it seems package_version (env$sess$request$httpuv.version) is the problem:

> lobstr::ref(package_version("0.0.1"))
Error in is.character(lhs) : node stack overflow

[hadley]

@yutannihilation do you want to have a look at why this is a problem? I suspect it's because:

x <- package_version("0.0.1")
identical(x, x[[1]])
#> [1] TRUE

^{Created on 2019-01-19 by the reprex package (v0.2.1.9000)}

So we need to use .subset2() instead of [[

[yutannihilation]

Yes, I'd like to (but maybe not today...).

[hadley]

No problems, the most important thing is the reprex 😄

[yutannihilation]

It seems you are right; ref_tree() is applied infinitely here because lapply(<package_version>, f) passes the newly generated package_version, not its elements, to f().

lobstr/R/ref.R

Line 57 in 9fcf642

subtrees <- lapply(x, ref_tree, layout = layout, seen = seen, character = character)

The problem is [[.numeric_version() creates a new object, so it never matches the addresses in seen.

library(lobstr)

x <- package_version("1.0.0")

obj_addr(x)
#> [1] "0x156f7138"

obj_addr(x[[1]])
#> [1] "0x17f49378"

obj_addr(x[[1]])
#> [1] "0x17f6d648"

^{Created on 2019-01-25 by the reprex package (v0.2.1)}

Using .subset2() might be an idea, but I think it's not always possible to know the "correct" length of the object to generate indices, because length() might tell us some wierd value.

Instead, how about unclass() in all these recursive cases? Since we are looking into the underlying structure, we don't need custom classes and methods. This means we can avoid calling such methods as as.list.reactiveValues(), so I feel this fix is more general.

lobstr/R/ref.R

Lines 55 to 62 in 9fcf642

	# recursive case
	if (is.list(x)) {
	subtrees <- lapply(x, ref_tree, layout = layout, seen = seen, character = character)
	} else if (is.environment(x)) {
	subtrees <- lapply(as.list(x), ref_tree, layout = layout, seen = seen, character = character)
	} else if (is.character(x)) {
	subtrees <- ref_tree_chr(x, layout = layout, seen = seen)
	}

If this looks OK, I'll create a PR.

[hadley]

Yeah, that sounds right to me.

[yutannihilation]

Thanks, I created #36.

BTW, I didn't know environments cannot be unclassed...

e <- structure(new.env(), class = "foo")
class(e)
#> [1] "foo"
unclass(e)
#> Error in unclass(e): cannot unclass an environment

^{Created on 2019-01-26 by the reprex package (v0.2.1)}