Inconsistent usage of `.by` vs. `by` across different dplyr verbs

[deschen1]

mutate, filter, and others use .by for grouping just within the respective verb.
slice_XXX (and maybe others) use by for grouping.

This is inconsistent and it would be great if this could be aligned across dplyr (if not the whole tidyverse).

[DavisVaughan]

While inconvenient, it is purposeful.

https://dplyr.tidyverse.org/reference/dplyr_by.html?q=by#supported-verbs

Note that some dplyr verbs use by while others use .by. This is a purely technical difference.

mutate() takes named ..., so we use a leading . to avoid collisions with column names a user might want. For example, a user may want a column named keep =, so we use .keep to avoid interfering.

mutate(
  .data,
  ...,
  .by = NULL,
  .keep = c("all", "used", "unused", "none"),
  .before = NULL,
  .after = NULL
)

slice_head() and friends don't use ... at all, so we don't need the leading .

slice_head(.data, ..., n, prop, by = NULL)

There are a few historical accidents, such as .data in slice_head() which could have been just data, but we have thought about this in great detail.

We even have code that checks for a by vs .by typo where possible

dplyr/R/by.R

Lines 141 to 175 in be3e3a0

	check_by_typo <- function(..., by = NULL, error_call = caller_env()) {
	check_by_typo_impl(
	wrong = "by",
	right = ".by",
	by = {{ by }},
	error_call = error_call
	)
	}
	check_dot_by_typo <- function(..., .by = NULL, error_call = caller_env()) {
	check_by_typo_impl(
	wrong = ".by",
	right = "by",
	by = {{ .by }},
	error_call = error_call
	)
	}
	check_by_typo_impl <- function(
	wrong,
	right,
	by = NULL,
	error_call = caller_env()
	) {
	by <- enquo(by)
	if (quo_is_null(by)) {
	return(invisible())
	}
	message <- c(
	"Can't specify an argument named {.code {wrong}} in this verb.",
	i = "Did you mean to use {.code {right}} instead?"
	)
	cli::cli_abort(message, call = error_call)
	}