Improve the error message for wrongly-placed :=

[HughParsonage]

As an experienced data.table user, I can generally recognize the cause of the := error message

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

However, (a) it still takes a bit of work, and (b) it doesn't fully diagnose the problem, even though it's possible. The following is a proof-of-concept that provides slightly better advice for the common error of providing an update expression to i = instead of j = .

Obviously the performance is irrelevant as it is only going to error at this point.

library(data.table)
DT <- as.data.table(mtcars)
DT[mpg_per_cyl := mpg/cyl]
#> Error: Check that is.data.table(DT) == TRUE. Otherwise, := and `:=`(...) are defined for use in j, once only and in particular ways. See help(":=").

":=" <- function(x, y) {
  .trace_back <- rlang::trace_back()
  if (length(.trace_back) >= 2L && 
      length(.trace_back[[2]]) >= 3L && 
      as.character(.trace_back[[1L]][[2]][[3]][[1]]) == ":=") {
    stop(":= is intended for use in j only, but is present in i. ", 
         "(Did you forget to put a comma ",
         "before the expression?)")
  } else {
    stop("Check that is.data.table(DT) == TRUE. Otherwise, := and `:=`(...) are defined for use in j, once only and in particular ways. See help(\":=\").")
  }
}

DT[mpg_per_cyl := mpg/cyl]
#> Error: := is intended for use in j only, but is present in i. (Did you forget to put a comma before the expression?)

^{Created on 2020-02-03 by the reprex package (v0.3.0)}