Dplyr: not detecting variables use with .env in mutate

[latot]

Hi!, well, this is weird, seems lintr is not detecting when var are used inside mutate with .env, I have others .env in filter and works fine, the code is from examples of how to use roads networks in R.

  source_nodes <- nodes %>%
    dplyr::filter(.data$start_end == "start") %>%
    dplyr::pull(.data$nodeID)

  target_nodes <- nodes %>%
    dplyr::filter(.data$start_end == "end") %>%
    dplyr::pull(.data$nodeID)

  edges <- edges %>%
    dplyr::mutate(from = .env$source_nodes, to = .env$target_nodes)

Logs:

warning: [object_usage_linter] local variable ‘source_nodes’ assigned but may not be used
  source_nodes <- nodes %>%
  ^~~~~~~~~~~~
warning: [object_usage_linter] local variable ‘target_nodes’ assigned but may not be used
  target_nodes <- nodes %>%

Thx!

[AshesITR]

If you remove .env, it should work fine. That construct is only necessary if there are name clashes between .data and .env.

[IndrajeetPatil]

For further details, see: https://adv-r.hadley.nz/evaluation.html#quoting-and-ambiguity

[latot]

Hi, remember this all are examples, the problem it self, is the function is not detecting if a var is used or not, maybe for that particular case there is no problem and we can apply workaround removing the .env, but in other code that this will affects will not be the case, like the next one:

This code have the same result as the first one.

library(magrittr)
library(rlang)

f <- function(nodes, edges) {
  a <- nodes %>%
    filter(.data$n == 2)
  b <- nodes %>%
    filter(.data$n == 3)
  return(dplyr::mutate(
    edges,
    from = .data$a * .env$a,
    to = .data$b * .env$b))
}

Thx!