bind_cols() treating data frame with 0 columns as NULL, i.e. not contributing

[romainfrancois]

closes #5300. This is labelled as a documentation issu in #5300 but I don't think it hurts to ignore 0 columns data frames.

library(dplyr, warn.conflicts = FALSE)

df <- data.frame(x = 1:3)
bind_cols(df, data.frame())
#>   x
#> 1 1
#> 2 2
#> 3 3

^{Created on 2020-06-29 by the reprex package (v0.3.0.9001)}

[romainfrancois]

Actually that does not seem to be a good idea, for the edge case when there are only 0 columns data frames, as you then lose the number of rows:

library(dplyr, warn.conflicts = FALSE)

df <- vctrs::new_data_frame(list(), n = 5L)
bind_cols(df)
#> # A tibble: 0 x 0

^{Created on 2020-06-29 by the reprex package (v0.3.0.9001)}

[lionel-]

I think we should be consistent with the recycling rules.