3

I have a moved a function I created into an R package and it has stopped working. I get an error:

Error in `:=`((value), 1) : 
Check that is.data.table(DT) == TRUE. Otherwise, := and `:=`(...) are defined for use in j, once only and in particular ways. See help(":=").

Weirdly this works if I load the function into the environment but not from the package? it seems very weird behaviour... I even check for whether it is a data.table in the function to stop this very thing!


data = as.data.table(mtcars)

# I Go into the package, highlight the function and run the code, to load the function into the environment  
# This works
create_dummy_var(data, var="cyl")

# This doesn't work (get the error message from above)
myPackage::create_dummy_var(data, var="cyl")

The original whole function

#' Converts a variable into a dummy variable
#'
#' Converts a single variable which contains a limited number of values into numeous dummary variables,
#' of 1 if the value is present 0 otherwise. The number of createds vars is equal to the number of unique observations in the varaiable
#'
#' @param dtable A data.table (must be data.table NOT data.frame or tibble)
#' @param var The variable to operate on
#'
#' @return data.table
#' @export
#'
#' @examples
#' 
#' df = data.frame("alp" = c("a", "a", "b", "d", "b"), "adults" = 1:5)
#' 
#' dt = data.table::data.table(df)
#' 
#' create_dummy_var(dt, "alp")
create_dummy_var <- function(dtable, var) {
  
  if (!var %in% names(dtable)) stop(var, " not found in dt")
  if(all(class(dtable) != "data.table")) stop("dt must be a data.table")
  
  dtable[, value := 1L]
  
  dcast(dtable , ...  ~ dtable[[var]] , value.var = "value", fill = 0, with = FALSE)
}

To replicate put this into an R package (you can use RStudio's new project, New Package template for the boiler plate).

> packageVersion("data.table")
[1] ‘1.14.2’
Joshua
  • 184
  • 1
  • 11

1 Answers1

4

Thanks to jangorecki for pointing out the Importing data.table vignette

The issue was declaring data.table's special symbols in the NAMESPACE.

The Importing data.table vignette does not mention that if you are using roxygen2 to generate the NAMESPACE then you can't use import(data.table) in the NAMESPACE. But as always the excellent usethis package has it covered, with usethis::use_data_table(). This creates all the boilerplate and it now works :)

Joshua
  • 184
  • 1
  • 11
  • I have a similar issue: `no visible global function definition for :=`. data.table is added in the DESCRIPTION file (but under Suggests, and not import). I also added `#' @import data.table` at the top of my R file. Could I ask where/how did you use use `this::use_data_table()`? – mOna Mar 28 '23 at 02:05
  • In the package root run `usethis::use_data_table()` . This will add a `R/datatabletest-package.R'` file - datatabletest being the name of my example package. It will also edit the DESCRIPTION and NAMESPACE files. - You need then `usethis` package installed – Joshua Mar 28 '23 at 09:39