Romain François

Thanks, I'll have a look.

I believe this should should be 1.0.8, and then we can think of something more intentional for 1.1.0: inform or warn about unknown attributes, hint about creating a custom class...

🤔 yeah perhaps as a follow up to this, `dplyr_row_slice.*` should all call `dplyr_reconstruct_core()` on the result of `vec_slice()` and then reconstruct the `groups`. But otherwise, I don't mind waiting...

This would be stricter than the current behavior that is somewhat forgiving: ``` r library(dplyr, warn.conflicts = FALSE) data.frame(x = 1:4) %>% slice(3:10) #> x #> 1 3 #> 2...

IIRC the oob forgiveness is for things like `slice(df, 1:5)` to get the first 5 rows in each group of `df` and be fine in groups where there are fewer...

Also not that trivial to implement. We can only realistically do that when R says the object has only one reference. This, to me, looks like modify by reference, à...

I see, I think I've been confused by the `group_by()`. see also `mutate_when()` from https://gist.github.com/romainfrancois/eeeed972d6734bcad3ec3dcf872df7ea ``` r library(rlang) library(dplyr) library(purrr) mutate_when x y #> #> 1 -1 -1 #> 2...

Here are some approaches using data frame returns: - "manually": ``` r library(dplyr) d % mutate({ test #> 1 -1 -1 #> 2 -2 -2 #> 3 -3 -3 #>...

Isn't this [with_groups()](https://dplyr.tidyverse.org/reference/with_groups.html) ?

Is this still needed @lionel- ?