Better document downside of named elements in `list_cbind()`

[Richard Ressler]

list_cbind calls vec_cbind which produces "packed data frame columns" with named inputs.

I am not familiar with packed data frame columns so was expecting to get a data frame with columns that were atomic vectors. The example from the list_cbind help (below) shows a different structure. This different structure also means the result of list_cbind displays differently in an R script than in a qmd/rmd code chunk .

x2 <- list(
  a = data.frame(x = 1:2),
  b = data.frame(y = "a")
)
list_cbind(x2) 
str(list_cbind(x2)) 
list_cbind(x2) |> unpack(cols = everything()) |> str()

Suggest adding additional information or suggestions in the help (or a vignette) on using unpack() how to convert the result of list_cbind into an unpacked data frame.

An alternative might be to add unpack to the list_cbind() as an argument to allow users to unpack as part of the function.

[Hadley Wickham]

What are you trying to do? It's most likely that you should avoid the packing step in the first place.

[Richard Ressler]

Thanks for responding (and everything else you do)!

I am not sure how to avoid the packing as

1. list_cbind() combines elements into a data frame by column-binding them together with vctrs::vec_cbind() and then

2. vec_cbind() creates packed data frame columns with named inputs.

I have not found a source that explains packed data frames or how to pass arguments to vec_cbind to unpack them as part of the functional call to list_cbind. Perhaps a vignette could be added to tidyr.

We can just pipe to flatten I guess but that seems to be an extra step that the newer functions such as pivot_wider etc try to avoid by using arguments.

Appreciate any recommendations.
Thanks,
Richard

[Hadley Wickham]

Oooh sorry, I think I missed the underlying issue here. You're getting this behaviour because list_cbind() is attempting to preserve the internal and external names. You can get the behaviour you want by stripping the names:

library(purrr)

x2 <- list(
  a = data.frame(x = 1:2),
  b = data.frame(y = "a")
)
str(list_cbind(unname(x2)))
#> 'data.frame':    2 obs. of  2 variables:
#>  $ x: int  1 2
#>  $ y: chr  "a" "a"

^{Created on 2023-07-27 with reprex v2.0.2}

I'll think about how to point this out in the docs.

[Richard Ressler]

Thanks for the explanation and the solution approach!

p.s. FYI even Chat GPT can't explain packed data.frames.:)

What is a good reference to explain the "packed data.frame" in R?
ChatGPT
As of my last update in September 2021, there isn't a native data structure called "packed data.frame" in R. ...