Data.table [, - c (...)] behaves differently if defined in a package

UPD. Thanks to @RYoda, I discovered that I forgot to include import(data.table)

in my NAMESPACE

file. However, I still don't understand why all other functions data.table

work fine even without it import()

, would be grateful if someone can explain it.

In my package, I define a function that, among other things, excludes certain columns from the data table. I can define it in different ways:

# code in the package: \testexcl\R\test_exclude.R

test_exclude_col1 <- function(inpDT,col2excl){
  output <- inpDT[, -c(col2excl), with=F]
  invisible(output)
}

test_exclude_col2 <- function(inpDT,col2excl){
  output <- inpDT[, setdiff(names(inpDT), col2excl), with=F]
  invisible(output;
}

      

        
        
        
      

    

Then I can use a function like this:

require(data.table)
require(testexcl) # my package
dt.iris <- data.table(iris)
dt.1 <- test_exclude_col1(dt.iris, 'Species') # Error
dt.2 <- test_exclude_col2(dt.iris, 'Species') # OK

      

        
        
        
      

    

test_exclude_col2()

works as expected but test_exclude_col1()

throws an error Error in -c(col2excl) : invalid argument to unary operator

. My biggest concern is that if I define exactly the same code test_exclude_col1()

not in the package, but in my session, it works correctly. Why is this happening?

I'm on R version 3.3.3 (2017-03-06), Win10 x86_64-w64-mingw32 / x64, data.table_1.10.4, tools_3.3.3

PS To make sure I am working with data.table, I added this part to the package:

test_exclude_col1a <- function(inpDT,col2excl){
  print(class(inpDT));
  inpDT <- data.table(inpDT);
  output <- inpDT[, -c(col2excl), with=F]
  invisible(output);
}

      

        
        
        
      

    

When I call test_exclude_col1a(dt.iris, 'Species')

it prints [1] "data.table" "data.frame"

, then gives the same error.