Can't pass additional arguments to group_modify() function

[earcanal]

I have some code which seems to have broken since group_map() became group_modify(). Converting to group_modify() only partially resolves this as my function also requires additional arguments. Under the old group_map() my function received args passed in ..., but I get an error for the same function under the new group_modify().

The docs indicate I should be able to do this:

... Additional arguments passed on to .f

Is this a regression, or am I doing something wrong?

# dplyr_0.8.3
library(tidyverse)

m <- function(x, y, a, b) {
  tibble(z = z)  
}

mtcars %>%
  group_by(cyl) %>%
  group_modify(m, 1, 2) %>%
  ungroup

Error in (function (.x, .y)  : unused arguments (1, 2)

[cderv]

I think there is indeed an issue in group_modify

dplyr/R/group_map.R

Lines 153 to 166 in ec09492

	fun <- function(.x, .y){
	res <- .f(.x, .y, ...)
	if (!inherits(res, "data.frame")) {
	abort("The result of .f should be a data frame")
	}
	if (any(bad <- names(res) %in% tbl_group_vars)) {
	abort(sprintf(
	"The returned data frame cannot contain the original grouping variables : ",
	paste(names(res)[bad], collapse = ", ")
	))
	}
	bind_cols(.y[rep(1L, nrow(res)), , drop = FALSE], res)
	}
	chunks <- group_map(.tbl, fun, ..., keep = keep)

the ... are passed to a new fun of two arguments .x and .y but then a re-pass in group_map applied on fun and this is where the error comes from because fun does not expect more than its two arguments.

I don't think the dots need to be passed in the second case and it should be

chunks <- group_map(.tbl, fun, keep = keep)

[billdenney]

I experienced the same issue just now. @cderv, I came up with the same fix before coming to the issue, so I included both of our names in the NEWS file.

[cderv]

Oh thanks ! I forgot about that one and forgot to do a PR 🤦‍♂
Thank @billdenney ! it was not necessary to mention me in NEWS but I appreciate it. 👍