group_by(.dots) handling - breaking change?

[jangorecki]

Code that used to work in current CRAN release 0.8.3, does not work anymore in recent devel.
Clean R environment, just installed cran remotes and github dplyr, as of 30m ago.
There seems to be a breaking change in the way how .dots argument is processed(?). I haven't found any references in NEWS file regarding this.

suppressMessages(library(dplyr))
data.frame(a=1:2, b=2:3) %>% group_by(.dots = c("a","b")) %>% summarize(count = n())
#Error: The quosure environment should be explicitly supplied as `env`
#Run `rlang::last_error()` to see where the error occurred.
rlang::last_error()
#<error/rlang_error>
#The quosure environment should be explicitly supplied as `env`
#Backtrace:
#  1. base::data.frame(a = 1:2, b = 2:3)
#  9. dplyr::group_by(., .dots = c("a", "b"))
# 10. dplyr::group_by_prepare(.data, ..., .add = .add)
# 11. dplyr:::compat_lazy_dots(.dots)
# 12. dplyr:::compat_lazy(dots[[i]], env, warn)
# 13. rlang::parse_quo(lazy[[1]], env)
#Run `rlang::last_trace()` to see the full context.
rlang::last_trace()
#<error/rlang_error>
#The quosure environment should be explicitly supplied as `env`
#Backtrace:
#     █
#  1. └─`%>%`(...)
#  2.   ├─base::withVisible(eval(quote(`_fseq`(`_lhs`)), env, env))
#  3.   └─base::eval(quote(`_fseq`(`_lhs`)), env, env)
#  4.     └─base::eval(quote(`_fseq`(`_lhs`)), env, env)
#  5.       └─`_fseq`(`_lhs`)
#  6.         └─magrittr::freduce(value, `_function_list`)
#  7.           └─function_list[[i]](value)
#  8.             ├─dplyr::group_by(., .dots = c("a", "b"))
#  9.             └─dplyr:::group_by.data.frame(., .dots = c("a", "b"))
# 10.               └─dplyr::group_by_prepare(.data, ..., .add = .add)
# 11.                 └─dplyr:::compat_lazy_dots(.dots)
# 12.                   └─dplyr:::compat_lazy(dots[[i]], env, warn)
# 13.                     └─rlang::parse_quo(lazy[[1]], env)

[hadley]

Thanks for the bug report — I've fixed the bug and I'm about to make it clear that this argument is deprecated. I'll do a soft deprecation which means it shouldn't affect user group, but it will encourage downstream packages (like dbplyr!) to not use it; we'll move to a stronger deprecation at some point in the future.

[tylerlittlefield]

I'm updating a package to remove usage of .dots the news file suggests !!!. I was just curious if that is still the case or if across might be preferred. I'm trying to decide which of the two examples I should use:

library(dplyr)
library(rlang)

# leaning towards this one
mtcars %>% 
  group_by(across(c("mpg", "cyl")))

mtcars %>% 
  group_by(!!!syms(c("mpg", "cyl")))

[hadley]

@tyluRp either is fine.