elephantupthewrongtree

You can also use `row_number()` in `slice()` ```r x %>% arrange(age) %>% group_by(age_group) %>% slice(head(row_number(), 2), tail(row_number(), 2)) ```

Would this require a change to `duckdb$rel_to_altrep`? That seems to be the roadblock. But maybe I'm not approaching it the right way. You can get it with DBI, but there's...

Yes, thanks for clarifying that code should already work. After upgrading everything including the R version, it's working on my end. FYI even with the latest package versions it did...

It works for me in 4.5, but not 4.3 reprex ``` r library(tidyverse) library(duckplyr) #> The duckplyr package is configured to fall back to dplyr when it encounters an #>...

This seems to work, passing all the tidyr nest tests (mostly, see notes below). Most of the work is already done just by using tidyr:::nest_info(). It doesn't handle any fall-back....

Sorry, I'm not following what you mean by `dd$` call. There are `dd::list()` and `dd::struct_pack()` calls in the `cols` expression being injected, so I assume you don't mean like `dd$list()`?...

Since duckplyr already imports rlang, tidyselect, and vctrs there's no new imports, it would just be a adding `nest_info()` and few small tidyr-internal functions used by `nest_info()`, about 100 lines...