Make pivot_wider error messages more useful, like spread errors

[hrecht]

Hello, I have found that the pivot_ family error messages are much less helpful than spread and gather error messages. For example, here's an error message from the exact same dataset that had a few rows with repeated values, using pivot_wider vs spread.

The spread error message tells me exactly where the issue is. The pivot_wider one is very difficult to understand and doesn't point me to the error. When I added values_fn = length to the pivot call I got another error, Error: 1 components of ...were not used. We detected these problematic arguments: *vaues_fn``
Could you please consider incorporating some of the old context?

pivot_wider:

Warning message:
Values are not uniquely identified; output will contain list-cols.
* Use `values_fn = list` to suppress this warning.
* Use `values_fn = length` to identify where the duplicates arise
* Use `values_fn = {summary_fun}` to summarise duplicates 

spread:

Error: Each row of output must be identified by a unique combination of keys.
Keys are shared for 6 rows:
* 427, 428
* 540, 541
* 800, 802

(edited to add text for error messages)

[hadley]

From my experience, most people found that error message cryptic; I don't think we want to recreate it.

[hrecht]

Understood. Could the new error messages be changed to something new then, ideally pointing out where the issues are? I find the current message very confusing. I've tried this suggestion - Use values_fn = length to identify where the duplicates arise - but that always gives another error. Perhaps it needs more description of how to use that argument.

[hadley]

Can you please provide a reprex?

[hrecht]

Here's an example. I couldn't quickly reproduce the length error on this dataset. But the output with values_fn = length working doesn't really show where the duplicate is. I'm not sure how that resulting tibble is helpful in identifying duplicates, especially if the dataset had thousands of rows.

library(tidyverse)

# Add a duplicate row
row_dupe <- fish_encounters %>% filter(row_number() == 20)
fish_dupe <- bind_rows(fish_encounters, row_dupe)

temp <- fish_dupe %>%
    pivot_wider(names_from = station, values_from = seen)
#> Warning: Values are not uniquely identified; output will contain list-cols.
#> * Use `values_fn = list` to suppress this warning.
#> * Use `values_fn = length` to identify where the duplicates arise
#> * Use `values_fn = {summary_fun}` to summarise duplicates

# Trying to follow suggestion in error message
fish_dupe %>%
    pivot_wider(names_from = station, values_from = seen, values_fn = length)
#> # A tibble: 19 x 12
#>    fish  Release I80_1 Lisbon  Rstr Base_TD   BCE   BCW  BCE2  BCW2   MAE   MAW
#>    <fct>   <int> <int>  <int> <int>   <int> <int> <int> <int> <int> <int> <int>
#>  1 4842        1     1      1     1       1     1     1     1     1     1     1
#>  2 4843        1     1      1     1       1     1     1     1     2     1     1
#>  3 4844        1     1      1     1       1     1     1     1     1     1     1
#>  4 4845        1     1      1     1       1    NA    NA    NA    NA    NA    NA
#>  5 4847        1     1      1    NA      NA    NA    NA    NA    NA    NA    NA
#>  6 4848        1     1      1     1      NA    NA    NA    NA    NA    NA    NA
#>  7 4849        1     1     NA    NA      NA    NA    NA    NA    NA    NA    NA
#>  8 4850        1     1     NA     1       1     1     1    NA    NA    NA    NA
#>  9 4851        1     1     NA    NA      NA    NA    NA    NA    NA    NA    NA
#> 10 4854        1     1     NA    NA      NA    NA    NA    NA    NA    NA    NA
#> 11 4855        1     1      1     1       1    NA    NA    NA    NA    NA    NA
#> 12 4857        1     1      1     1       1     1     1     1     1    NA    NA
#> 13 4858        1     1      1     1       1     1     1     1     1     1     1
#> 14 4859        1     1      1     1       1    NA    NA    NA    NA    NA    NA
#> 15 4861        1     1      1     1       1     1     1     1     1     1     1
#> 16 4862        1     1      1     1       1     1     1     1     1    NA    NA
#> 17 4863        1     1     NA    NA      NA    NA    NA    NA    NA    NA    NA
#> 18 4864        1     1     NA    NA      NA    NA    NA    NA    NA    NA    NA
#> 19 4865        1     1      1    NA      NA    NA    NA    NA    NA    NA    NA

^{Created on 2021-04-22 by the reprex package (v2.0.0)}

[hadley]

Hmmmm, you're looking for the cell with a 2 in it, but obviously that might be hard to find. What I would do in this case is:

library(tidyverse)

row_dupe <- fish_encounters %>% filter(row_number() == 20)
fish_dupe <- bind_rows(fish_encounters, row_dupe)

fish_dupe %>%
  count(fish, station, seen) %>% 
  filter(n > 1)
#> # A tibble: 1 x 4
#>   fish  station  seen     n
#>   <fct> <fct>   <int> <int>
#> 1 4843  BCW2        1     2

^{Created on 2021-04-22 by the reprex package (v2.0.0)}

So maybe the error message could suggest that more directly.