Provide a convenient way to silence "New names" message when using col_names = FALSE

[wgrundlingh]

When using col_names = FALSE, an automated message with "New names" lists the (new) column names. Can a feature be added that avoids printing this information? Something like messages = FALSE. It would be helpful when working with fixed/known inputs for which new column names are specified programmatically.

The following code replicates the display:

read_excel(path = readxl_example('clippy.xlsx'), col_names = FALSE)

Here is the output:

[jennybc]

readxl affords no control over this and I'm not aware of a tibble option that helps either.

It is ultimately emitted via message() from within the tibble package. My best advice is to muffle it with suppressMessages() like so:

library(readxl)

suppressMessages(
  read_excel(readxl_example("clippy.xlsx"), col_names = FALSE)
)
#> # A tibble: 5 x 2
#>   ...1                 ...2     
#>   <chr>                <chr>    
#> 1 name                 value    
#> 2 Name                 Clippy   
#> 3 Species              paperclip
#> 4 Approx date of death 39083    
#> 5 Weight in grams      0.9

^{Created on 2019-08-08 by the reprex package (v0.3.0.9000)}

[wgrundlingh]

My suggestion was to add this as a feature. Here is a simple implementation how I'm currently implementing suppressMessages(); not ideal:

read_excel__ <- read_excel

read_excel <- function (suppress_messages = FALSE, ...) {
  if (suppress_messages) {
    Suppress_Messages <- suppressMessages
  } else {
    Suppress_Messages <- function (args) { args } # NoOp
  }

  return (
    Suppress_Messages(
      read_excel__( ... )
    )
  )  
}

Is this something that seems a better fit with tibble (and therefore would be a better fit here)?

[jennybc]

Yeah I do think it's a better issue to ponder in tibble (or conceivably even further upstream). Because ... I'm not going to add an argument to read_excel() for this. So if it happened here, it would require a new wrapper function (what you've made and also a nonstarter) or a readxl option (readxl has no options, so also very unattractive). And if it's an option, it should probably be a tibble option about whether it's chatty about name repair.

[jennybc]

Also: what is your context? Is this about an Rmd document or similar?

[jennybc]

My reason for asking about .Rmd (or spinning .R) is that you have a really nice way to achieve your goals with the chunk option message = FALSE. Applied to individual chunks or all chunks.

[slyrus]

I get that I'm a couple years late to the party but I second the notion that this is a problem with readxl, not with generic tibble construction. IMO, readxl should provide default column names when none or provided - this shouldn't require changing the core tibble API/contract and readxl is in a much better position to understand how many columns are being read than the user code calling it. You could even use col_names = NULL as the flag to provide default colnames (e.g. "Col1" or "ColA") without changing the current behavior.

[sanjmeh]

New names:
• -> `...3` • -> ...4
• -> `...5` • -> ...6
• -> `...7` • -> ...8
• -> `...9` • -> ...10
• -> `...11` • -> ...12
• -> `...13` • -> ...14
• -> `...15` • -> ...16
• -> `...17` • -> ...18
• -> `...19` • -> ...20

This type of output is not at all helpful.
Why can read_excel() not have an option of suppress names message?

[xaviescacs]

In my case, I want to skip the first column and I indicate it through col_types with the "_" character which means "skip column". Yet I got the message that the name of this column has been changed, which I couldn't care less about. If I use cols_only() I get the same output. Perhaps is this case It would be reasonable to don't show those messages. I don't want to use suppressMessages() because I don't want to hide any other potentially meaningful messages.

[bhollan]

I can confirm this is still happening and still annoying.

[RyanWilkins]

Would also like a solution to this, please

[jennybc]

Thanks to a change in a low-level package (vctrs), we now have a really nice solution to this in readxl and elsewhere. Add .name_repair = "unique_quiet" to the call.

library(readxl)

packageVersion("vctrs")
#> [1] '0.5.1'

read_excel(path = readxl_example('clippy.xlsx'), col_names = FALSE)
#> New names:
#> • `` -> `...1`
#> • `` -> `...2`
#> # A tibble: 5 × 2
#>   ...1                 ...2     
#>   <chr>                <chr>    
#> 1 name                 value    
#> 2 Name                 Clippy   
#> 3 Species              paperclip
#> 4 Approx date of death 39083    
#> 5 Weight in grams      0.9

read_excel(
  path = readxl_example('clippy.xlsx'),
  col_names = FALSE,
  .name_repair = "unique_quiet"
)
#> # A tibble: 5 × 2
#>   ...1                 ...2     
#>   <chr>                <chr>    
#> 1 name                 value    
#> 2 Name                 Clippy   
#> 3 Species              paperclip
#> 4 Approx date of death 39083    
#> 5 Weight in grams      0.9

^{Created on 2022-12-09 with reprex v2.0.2.9000}

Place to read more if you're curious: r-lib/vctrs#1629