No way to remove duplicate column names

[jzadra]

Some joins will allow a tibble to end up with two columns with identical names. Doing any sort of tidyselect operation on the tibble results in:

Error: Can't bind data because some arguments have the same name

However, there doesn't appear to be any way to remove the duplicated column. It seems to me that using select(-matches("duplicate name")) or select(-contains("duplicated name")) ought to work since they allow catching multiples. In addition, selecting based on the index should also work. If columns 100 and 101 have the same name, I should be able to select(-101) however these all result in the same Error: Can't bind data because some arguments have the same name

[moodymudskipper]

I have already encountered this issue too. I believe the logic is that dplyr wants to enforce tidy data so you need to clean it first using tibble or tidyr, and then use dplyr functions, but the tidyverse only proposes tibble::repair_names() to deal with duplicate names as far as I can tell, which sometime feels clunky. Maybe an option .name_repair = "drop" to tibble::as_tibble() would help ?.

I think in any case it would be helpful to add a suggestion for a workaround to the error message select triggers.

Here's a workaround :

packageVersion("dplyr")
#> [1] '0.8.3'
library(dplyr, warn.conflicts = F)
test <- setNames(data.frame(a=1,b=2,b=3,c=4), c("a","b","b","c"))
test
#>   a b b c
#> 1 1 2 3 4
tibble::repair_names(test,sep= "***") %>%
  select(-contains("***"))
#>   a b c
#> 1 1 2 4

^{Created on 2019-11-15 by the reprex package (v0.3.0)}

[jzadra]

Thanks for that. My only concern about this as a long term solution would be the risk of affecting other names in the tibble unintentionally.

[moodymudskipper]

test[!duplicated(names(test))] is probably the safest way to go

[hadley]

We now have the tooling to handle this in the dev version of tidyselect and it should be straight forward to update dplyr.