Feature request: can data_syms() keep NULL elements as-is?

[teunbrand]

When passing variable names as string, is seems natural to have optional arguments default to NULL.
However, this makes it difficult to use data_syms(), as it cannot convert these.

library(rlang)

f <- function(df, a = NULL, b = NULL) {
  syms <- data_syms(list(a = a, b = b))
  vctrs::data_frame(!!!lapply(syms, eval_tidy, data = df))
}

mtcars |> f("cyl", "disp") |> head()
#>   a   b
#> 1 6 160
#> 2 6 160
#> 3 4 108
#> 4 6 258
#> 5 8 360
#> 6 6 225

# Here, we only want the `a` column
mtcars |> f("cyl") |> head()
#> Error in `sym()`:
#> ! Can't convert `NULL` to a symbol.

^{Created on 2024-12-09 with reprex v2.1.1}

My proposal is to make it easy to have data_syms(list(NULL)) return list(NULL).
For context, this is related to tidyverse/ggplot2#6208.

[lionel-]

hmm I don't know. It seems like it's better to be explicit at the call site by calling compact() before data_syms(), that will make it very clear what's going on and is still quite terse.

[teunbrand]

Thanks for the response Lionel. I think we'd like to preserve NULL rather than filter out NULL-elements. From ggplot2's end, geom_point(mapping = aes(colour = NULL)) negates a global mapping in ggplot(mapping = aes(colour = foo)). I think this is useful behaviour to carry over.