Check if supplying named arguments works with logged functions

[b-rodrigues]

Noticed that r_unique <- record(data.table:::unique.data.table)

then r_unique(a, by = c(x,y)) seems to ignore by argument

[b-rodrigues]

mode <- function(x) {
  ux <- unique(x)
  ux[which.max(tabulate(match(x, ux)))]
}

vetor <- c(3,1,2,5,9,3,2,2,5,8,4,1)
mode(x=vetor)

# decorate function
r_mode <- record(mode, .g = length, diff = 'none')

# this works
res <- r_mode(vetor)
read_log(res)
check_g(res)

# this doesn’t
r_mode(x=vetor)

# if the function has several arguments, we can name the second but not the first
soma2 <- function(x, y){
  res <- x+y
  
  return(res)
}

soma2(x=2, y=6)

# this works
r_soma2 <- record(soma2, .g = length, diff = 'none')

r_soma2(2,6)

r_soma2(2, y=6)

# this doesn’t
r_soma2(x=2,y=6)

@MatousEibich any ideas? I’ll try some things out this weekend

[MatousEibich]

@b-rodrigues Looked into it a little bit. I have a question related to this issue - why do we need .value (mainly in purely())? My understanding is that it's a placeholder for the first argument of the function being recorded. Can't we just use dynamic dots without .value and then change the code in purely to something like:

if(all(purrr::map_lgl(args, maybe::is_nothing))){

      final_result <- list(
        value = maybe::nothing(),
        log_df = "A `Nothing` was given as input."
      )

    } 

Is my understanding correct or am I missing something?

[b-rodrigues]

sorry, I'm suuuuuuper late, but seems you're right! I've started fixing this issue in the fixing_issue7 branch.

[b-rodrigues]

hmm actually keeping .value seems to be important, especially for functions with NSE (dplyr, mainly). Basically a decorated function has its first argument overwritten to .value, and I don't know how this could be avoided