# Why would I use this feature?

The main use case of this feature is to support traceability of package and function use within a validated environment.

Following the guidance outlined in A Risk-Based Approach for Assessing R Package Accuracy within a Validated Infrastructure, packages are selected to support use cases, risk is assessed and mitigated, and packages are included in your validated environment for use. Section 4.3 of this paper nicely outlines the need to trace what is used and the need to identify package and function use where risk was not assessed for direct use within the validated environment.

4.3. Traceability

“One of the core concepts presented in this paper is that Imports are not typically loaded by users and need not therefore be directly risk-assessed. If adopting this risk-based approach then measures need to be taken to ensure that users do not directly load the Package Imports. It is suggested that this is handled mainly through process, although tools could be developed to check using sessionInfo or devtools::session_info that check the loaded packages against packages lists of Intended for Use and Imports. In any case the use of these tools within a standard, logged, workflow is highly recommended to ensure traceability of the work.”

{logrx} provides you this tool! It even goes a step further by not just logging the packages you’ve use, but it will log use at a function level. This gives you the flexibility of assessing the risk of functions and approving functions, rather than having to assess the risk of the entire package for use within your validated environment.

# How do I use this feature?

## 1. Create a named list

The named list contains the functions approved for use for each package. If all functions for a package are approved for use, list “All”.

approved_pkgs <- list(base = "mean",
dplyr = "All")
approved_pkgs
#> $base #> [1] "mean" #> #>$dplyr
#> [1] "All"

## 2. Build approved.rds

Pass the named list through build_approved() to build your tibble. We create a temp directory to save this for illustration.

build_approved(approved_pkgs)
#> # A tibble: 289 × 2
#>    function_name library
#>    <chr>         <chr>
#>  1 mean          package:base
#>  2 rows_upsert   package:dplyr
#>  3 src_local     package:dplyr
#>  4 db_analyze    package:dplyr
#>  5 n_groups      package:dplyr
#>  6 distinct      package:dplyr
#>  7 summarise_    package:dplyr
#>  8 group_split   package:dplyr
#>  9 group_by_if   package:dplyr
#> 10 sample_n      package:dplyr
#> # … with 279 more rows

## 3. Save as approved.rds

You can use the file argument in build_approved() to save approved.rds instead of returning the tibble.

dir <- tempdir()

build_approved(approved_pkgs, file = file.path(dir, "approved.rds"))

## 4. Update the logrx.approved option

Update the logrx.approved option to point to your approved.rds location. We recommend setting this in your .Rprofile.

options(log.rx.approved = file.path(dir, "approved.rds"))

## 5. You’re done. Let’s axecute!

logrx will take it from there. When each program is executed, packages and functions will be compared to approved.rds and if any unapproved use is found, it will be logged within the “Unapproved Package and Functions” section of the log file.

# Example

Let’s write a simple script summarizing mean mpg from mtcars. We save this as mpg.R in the temporary directory dir and axecute() it.

library(dplyr, warn.conflicts = FALSE)

results <- mtcars %>%
group_by(cyl) %>%
summarize(mean = mean(mpg)) %>%
mutate(label = "Mean MPG")

results %>%
tidyr::pivot_wider(names_from = cyl, values_from = mean, id_cols = label)
#> # A tibble: 1 × 4
#>   label      4   6   8
#>   <chr>    <dbl> <dbl> <dbl>
#> 1 Mean MPG  26.7  19.7  15.1
axecute(file.path(dir,"mpg.R"), remove_log_object = FALSE)

Here we have the log elements for “Used Package and Functions” and “Unapproved Package and Functions”. We can see we used library() from package:base and pivot_wider from package:tidyr. We did not include the base library or tidyr functions in our approved list, so this has been logged!

#> --------------------------------------------------------------------------------
#> -                          Used Package and Functions                          -
#> --------------------------------------------------------------------------------
#> {package:base} library, mean
#> {package:dplyr} %>%, group_by, summarize, mutate
#> {package:tidyr} pivot_wider
#> --------------------------------------------------------------------------------
#> -                       Unapproved Package and Functions                       -
#> --------------------------------------------------------------------------------
#> {package:base} library
#> {package:tidyr} pivot_wider