Documentation: Mix of two Higher Order Functions

[yurovska]

Please select a category the issue is focused on?

Function Documentation

Let us know where something needs a refresh or put your idea here!

It is not really clear from documentation whether it's kinda "legitimate" to use one Higher Order function inside of another.

I mean, is that OK to do something like this?

advs <- advs %>%
  restrict_derivation(
    derivation = call_derivation,
    args = params(
      derivation = derive_vars_crit_flag,
      variable_params = list(
        params(
          condition = PCHG <= -5,
          description = "Body weight reduction ≥5% from baseline",
          crit_nr = 1
        ),
        params(
          condition = PCHG <= -10,
          description = "Body weight reduction ≥10% from baseline",
          crit_nr = 2
        ),
        params(
          condition = PCHG <= -15,
          description = "Body weight reduction ≥15% from baseline",
          crit_nr = 3
        ),
        params(
          condition = PCHG <= -20,
          description = "Body weight reduction ≥20% from baseline",
          crit_nr = 4
        ),
        params(
          condition = PCHG <= -25,
          description = "Body weight reduction ≥25% from baseline",
          crit_nr = 5
        )
      ),
      values_yn = TRUE,
      create_numeric_flag = FALSE
    ),
    filter = PARAMCD == "WGHT" & ONTRTFL == "Y"
  )

[bundfussr]

Yes, using higher order functions is fine because they fulfill the requirements for the derivation argument (first argument must expect a dataset and the function must return a dataset).

We should update the documentation and change

The function must provide the `dataset` argument and all arguments specified in the `params()` objects passed to the `arg` argument.

to

The function must provide all arguments specified in the `params()` objects passed to the `args` argument.

because the dataset argument is no longer expected.

@yurovska , what else would you suggest to update.

[yurovska]

@yurovska , what else would you suggest to update.

@bundfussr Thanks a lot for your reply!

I remember somewhere there is a note saying that {dplyr} functions like mutate() are not suitable for the derivation argument. I think it is worth also to mention explicitly that another Higher Order function is fine there, and add at least one example.

[bundfussr]

I remember somewhere there is a note saying that {dplyr} functions like mutate() are not suitable for the derivation argument.

The note was removed when we updated the checks such that the first argument doesn't need to be named dataset.