Possible cause for object '.x' not found error with tar_combine

[nicolaspayette]

I had a problem, for which I eventually found the cause, but it took me quite some time to figure it out, so I'm posting it here in case some else runs into the same issue.

Take the following example:

library(targets)
tar_script({
  library(tibble)
  library(tarchetypes)
  targets <- tar_map(
    values = lst(letter = c("A", "B", "C")),
    tar_target(x, tibble(value = letter))
  )
  lst(
    targets,
    tar_combine(y, targets, command = dplyr::bind_rows(!!!.x, .id = "name"))
  )
})

tar_manifest(fields = command)
#> Error:
#> ! Error running targets::tar_manifest()
#>   Target errors: targets::tar_meta(fields = error, complete_only = TRUE)
#>   Tips: https://books.ropensci.org/targets/debugging.html
#>   Last error: object '.x' not found

#> Backtrace:
#>     ▆
#>  1. └─targets::tar_manifest(fields = command)
#>  2.   └─targets:::callr_outer(...)
#>  3.     ├─targets:::if_any(...)
#>  4.     └─targets:::callr_error(condition = out, fun = fun)
#>  5.       └─targets::tar_throw_run(message)
#>  6.         └─targets::tar_error(...)
#>  7.           └─rlang::abort(message = message, class = class, call = tar_empty_envir)

^{Created on 2022-11-30 with reprex v2.0.2}

It fails with object '.x' not found. Can you see why?

The culprit is the tibble::lst function, which I have used here to build the target list. It doesn't work with the slicing operator !!!.x, presumably because it's doing something fancy under the hood.

Using plain old base list instead fixes the issue:

library(targets)
tar_script({
  library(tibble)
  library(tarchetypes)
  targets <- tar_map(
    values = lst(letter = c("A", "B", "C")),
    tar_target(x, tibble(value = letter))
  )
  list(
    targets,
    tar_combine(y, targets, command = dplyr::bind_rows(!!!.x, .id = "name"))
  )
})

tar_manifest(fields = command)
#> # A tibble: 4 × 2
#>   name  command                                                                 
#>   <chr> <chr>                                                                   
#> 1 x_B   "tibble(value = \"B\")"                                                 
#> 2 x_C   "tibble(value = \"C\")"                                                 
#> 3 x_A   "tibble(value = \"A\")"                                                 
#> 4 y     "dplyr::bind_rows(x_A = x_A, x_B = x_B, x_C = x_C, \n     .id = \"name\…

^{Created on 2022-11-30 with reprex v2.0.2}

I'm happy to carry on from here, but if someone could explain why lst doesn't play well with !!!.x, I'd be very curious.

[wlandau]

The problem is that lst() tries to tidy-evaluate .x. We instead want lst() to ignore .x and leave it for tar_combine(). This works:

targets <- tar_map(
  values = lst(letter = c("A", "B", "C")),
  tar_target(x, tibble(value = letter))
)
combined <- tar_combine(y, targets, command = dplyr::bind_rows(!!!.x, .id = "name"))
lst(targets, combined)