Feature Request: add .groups = "drop" option to mutate

[sbanville-delfi]

Currently summarise provides the .groups = "drop" parameter (along with other options), to eliminate the need to call ungroup. I think having this option available to mutate would also be helpful.

For example, it would be a useful shortcut to drop groups in mutate by doing the following:

library(ggplot2)
mpg |>
  dplyr::arrange(displ) |>
  dplyr::group_by(displ) |>
  dplyr::mutate(
    avg_hwy = mean(hwy),
    .groups = "drop"
  ) |>
  as.data.frame()

This isn't a high priority of course.

Thanks in advance.

[DavisVaughan]

I think this is much nicer

library(ggplot2)
mpg |>
  dplyr::arrange(displ) |>
  dplyr::mutate(
    avg_hwy = mean(hwy),
    .by = displ
  ) |>
  as.data.frame()

Same for summarise(), really, instead of having to specify .groups.

For summarise(), we only introduced .groups because we were already doing some "special things" with the groups before we returned them, and this gave the user a way to control that. For mutate(), if you pass in a data frame grouped by group_by() then there is only ever 1 behavior - it retains those groups. So I don't think there is any need to add the same kind of customization argument here. But thanks!

[sbanville-delfi]

I think this is much nicer

library(ggplot2)
mpg |>
  dplyr::arrange(displ) |>
  dplyr::mutate(
    avg_hwy = mean(hwy),
    .by = displ
  ) |>
  as.data.frame()

Same for summarise(), really, instead of having to specify .groups.

For summarise(), we only introduced .groups because we were already doing some "special things" with the groups before we returned them, and this gave the user a way to control that. For mutate(), if you pass in a data frame grouped by group_by() then there is only ever 1 behavior - it retains those groups. So I don't think there is any need to add the same kind of customization argument here. But thanks!

Thanks @DavisVaughan for the quick response!
Yes, agreed. I didn't realize this was doable already in mutate; now I know.

[DavisVaughan]

@sbanville-delfi you might enjoy reading https://www.tidyverse.org/blog/2023/02/dplyr-1-1-0-per-operation-grouping/