Stop using the library() function

Yes, i know, you love using the library() function in R! But it can be a real problem for several reasons:

• It makes your code too ambiguous
• It can mask other functions with similar names, if you have multiple library statements
• It is forbidden in R package creations

Let’s take each one of these points in turn

Ambiguous code

When you library multiple packages, it becomes unclear which functions belong to which packages. This makes your code really difficult to read for other users, especially if you request help

Masking/ name clashes

When you library multiple packages, if any of the packages contain functions with the same name, or contain functions with the same name as functions in packages bundled with R (like stats or utils) then all but one will be masked. The order in which packages are libraried matters. The most recent one will take precedence

Consider this example

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

^{Created on 2025-02-10 with reprex v2.1.0}

Just this one library command has masked a whole bunch of functions from base R.

Creating R packages

Eventually, you will evolve as an R coder to realizing that you want/need to write an R package. At this point, the above points will make more sense (if they do not already).

When writing R functions you are not permitted to use the library command in any function. You must, instead, reference every function using the package they come from (i.e their namespace). This is achieved using the :: operator.

For example

iris |> 
  tidyr::pivot_longer(cols = -Species,
                      names_to = "names",
                      values_to = "values") |> 
  dplyr::group_by(Species,names) |> 
  dplyr::summarise(mn=base::mean(values),
                   .groups = "drop")

Here, the pivot_longer function comes from the tidyr package, summarize and group_by come from the dplyr package and mean comes from base.

In conclusion: Yes, you do end up typing a lot more but overall it’s a good practice!