We can't use unnest_wider() on vctrs list columns

[moodymudskipper]

See #737 for full example

These are the columns that pivot_wider() creates so I don't believe this is expected :

library(tidyr)

df1 <- tibble::tribble(
  ~type,          ~name, ~var1,
  "Country",   "Norway", 169L, 
    "Sport",     "Skii", 169L, 
  "Country",    "Spain", 150L, 
    "Sport",     "Bike", 150L, 
    "Sport",   "Soccer", 150L, 
    "Sport",   "Basket", 150L, 
  "Country",      "USA",   0L, 
    "Sport", "Baseball",   0L, 
)

df2 <- pivot_wider(df1, names_from = "type", values_from = "name") 
#> Warning: Values in `name` are not uniquely identified; output will contain list-cols.
#> * Use `values_fn = list(name = list)` to suppress this warning.
#> * Use `values_fn = list(name = length)` to identify where the duplicates arise
#> * Use `values_fn = list(name = summary_fun)` to summarise duplicates
df2
#> # A tibble: 3 x 3
#>    var1     Country       Sport
#>   <int> <list<chr>> <list<chr>>
#> 1   169         [1]         [1]
#> 2   150         [1]         [3]
#> 3     0         [1]         [1]

unnest_wider(df2, Sport)
#> New names:
#> * `` -> ...1
#> New names:
#> * `` -> ...1
#> * `` -> ...2
#> * `` -> ...3
#> New names:
#> * `` -> ...1
#> No common type for `x` <tbl_df<...1:character>> and `y` <character>.

df2$Sport <- unclass(df2$Sport)

unnest_wider(df2, Sport)
#> New names:
#> * `` -> ...1
#> New names:
#> * `` -> ...1
#> * `` -> ...2
#> * `` -> ...3
#> New names:
#> * `` -> ...1
#> # A tibble: 3 x 5
#>    var1     Country ...1     ...2   ...3  
#>   <int> <list<chr>> <chr>    <chr>  <chr> 
#> 1   169         [1] Skii     <NA>   <NA>  
#> 2   150         [1] Bike     Soccer Basket
#> 3     0         [1] Baseball <NA>   <NA>

pivot_wider(df1, names_from = "type", values_from = "name", values_fn = list(name=list))
#> # A tibble: 3 x 3
#>    var1     Country       Sport
#>   <int> <list<chr>> <list<chr>>
#> 1   169         [1]         [1]
#> 2   150         [1]         [3]
#> 3     0         [1]         [1]

^{Created on 2019-09-15 by the reprex package (v0.3.0)}

[hadley]

Can you please work on making your reprexes more minimal? This code gets to the essence of the problem much more quickly:

library(tidyr)

df <- tibble(
  x = 1:2,
  y = vctrs::list_of(c("a", "b"), "c")
)
df %>% unnest_wider(y)
#> New names:
#> * `` -> ...1
#> * `` -> ...2
#> New names:
#> * `` -> ...1
#> No common type for `x` <tbl_df<
#>   ...1: character
#>   ...2: character
#> >> and `y` <character>.
df %>% unnest_longer(y)
#> No common type for `x` <tbl_df<y:character>> and `y` <character>.

^{Created on 2019-09-16 by the reprex package (v0.3.0)}

[DavisVaughan]

@hadley see the PR before continuing!