Cannot specify col_types for columns with missing names

[stephenbalogun]

I have a relatively heavy .csv file (over 110MB) that I routinely analyze and use the vroom() for data import. Recently, I decided to be explicit with the col_types parameter but vroom keeps throwing an error because one of the column names is missing. I noticed that this does not happen with the readr::read_csv() but it is significantly slower. Would be helpful to have this addressed also in vroom. Thank you.

library(vroom)
vroom("x,\n1,2\n3,4", delim = ",", col_types = cols(
    x = col_double(),
    `...2` = col_double()
))
#> New names:                                                                                                             
#> * NA -> ...2
#> Error: Invalid input type, expected 'list' actual 'NULL'
#> In addition: Warning message:
#> The following named parsers don't match the column names: ...2 

library(readr)
read_csv("x,\n1,2\n3,4", col_types = cols(
    x = col_double(),
    `X2` = col_double()
))

#> # A tibble: 2 x 2
#>     x    X2
#>  <dbl> <dbl>
#> 1     1     2
#> 2     3     4
#> Warning message:
#> Missing column names filled in: 'X2' [2] 

[jimhester]

Thanks for opening the issue and a reproducible example! I will see if we can fix this.

A workaround is to specify the types without names., e.g.

vroom::vroom("x,\n1,2\n3,4", delim = ",", col_types = "dd")
#> New names:
#> * `` -> ...2
#> # A tibble: 2 x 2
#>       x  ...2
#>   <dbl> <dbl>
#> 1     1     2
#> 2     3     4

^{Created on 2021-02-16 by the reprex package (v1.0.0)}

[stephenbalogun]

Thank you for the feedback. Your recommendation is exactly what I am currently doing. The only challenge with that is that the number of columns in my datafile keeps changing when new variables are added. In the original script, I used the cols_only() function with the col_types argument so my code does not break when newer variables are added e.g.

# this works
vroom("x,y\n1,2\n3,4", delim = ",", col_types = cols_only(
   x = col_double(),
   y = col_double()
))
#> # A tibble: 2 x 2                                                                                                           
#>      x     y
#>  <dbl> <dbl>
#> 1     1     2
#> 2     3     4


# when additional columns are added, it still works!
vroom("x,y,z\n1,2,3\n4,5,6", delim = ",", col_types = cols_only(
   x = col_double(),
   y = col_double()
))
#> # A tibble: 2 x 2                                                                                                           
#>      x     y
#>  <dbl> <dbl>
#> 1     1     2
#> 2     3     4

# unlike this when  the number of variables change
vroom("x,y,z\n1,2,3\n4,5,6", delim = ",", col_types = "dd")
#> Error: Unnamed `col_types` must have the same length as `col_names`. 

I hope that your team is able to fix this soon.

Again, thank you for providing an incredibly fast package for importing flat files that works well with other tidyverse packages.

[jimhester]

Thank you for opening the issue and for supplying a reproducible example, it is a big help!