Introduction

This vignette takes the user through the basic functionality of the review package. The user will gain experience with:

  • creating and updating a QC log
  • generating simple visual diffs to use during QC
  • generating QC summary documents

Setup

Load libraries

Create demo repo

demoRepo() creates a temporary SVN repo containing files with version history. This repo already contains a QC log and partially completed QC for demonstration purposes. The function returns the path to the repo.

repo <- demoRepo("abc-123")
print(repo)
## [1] "/tmp/Rtmphbxmvo/svn-proj/abc-123"

Using the QC log

If a repo does not contain a QC log, the first step would be to run logCreate(), which will generate an empty one. As mentioned above, the demo repo already contains a QC log. The contents are shown below.

## # A tibble: 7 × 6
##   file                   origin                  revf  revo reviewer time       
##   <chr>                  <chr>                  <dbl> <dbl> <chr>    <chr>      
## 1 script/data-assembly.R script/data-assembly.R     0     0 anyone   2024-06-04…
## 2 script/pk/load-spec.R  script/pk/load-spec.R      0     0 anyone   2024-06-04…
## 3 script/combine-da.R    script/combine-da.R        0     0 anyone   2024-06-04…
## 4 script/examp-txt.txt   script/examp-txt.txt       0     0 anyone   2024-06-04…
## 5 script/data-assembly.R script/data-assembly.R     1     1 michaelm 2024-06-04…
## 6 script/pk/load-spec.R  script/pk/load-spec.R      1     1 michaelm 2024-06-04…
## 7 script/combine-da.R    script/combine-da.R        1     1 michaelm 2024-06-04…

A few helpful definitions:

  • file: Relative file path from the location of the QC log.
  • origin: Relative file path of the origin file.
  • revf: the revision number of the file that was reviewed.
  • revo: the revision number of the origin file at time indicated.
  • reviewer: the reviewer of the file, given as the equivalent in R to Sys.info()[["user"]].
  • time: the time of the assignment or acceptance.

logAssign

When a file is ready for QC, it should be assigned in the QC log. This can be done using the logAssign() function, which will add a record to the QC log.

Looking at the demo QC log, the first row is an example of a record that logAssign() will generate.

## # A tibble: 1 × 6
##   file                   origin                  revf  revo reviewer time       
##   <chr>                  <chr>                  <dbl> <dbl> <chr>    <chr>      
## 1 script/data-assembly.R script/data-assembly.R     0     0 anyone   2024-06-04…

Notice that revf is 0 for all entries. Additionally, the reviewer is set to “anyone”. This assignment is done to signify the file is ready for QC, but revf = 0 indicates no QC has been completed yet.

If we want to assign another file, such as script/examp-yaml.yaml, we can do the following:

logAssign("script/examp-yaml.yaml")

logAccept

After the reviewer completes their QC of a file, logAccept() should be used to create a record of this. The revision number at the time of QC is captured along with the username of the reviewer.

We can look at the QC history of script/data-assembly.R to demonstrate the differences between logAssign() and logAccept().

## # A tibble: 2 × 6
##   file                   origin                  revf  revo reviewer time       
##   <chr>                  <chr>                  <dbl> <dbl> <chr>    <chr>      
## 1 script/data-assembly.R script/data-assembly.R     0     0 anyone   2024-06-04…
## 2 script/data-assembly.R script/data-assembly.R     1     1 michaelm 2024-06-04…

Notice that the row created by logAccept() updated the revf value to 1, which is the revision number where the QC was completed. The reviewer was updated from “anyone” to the actual user who did the QC.

QC log and SVN

It is important to note that after any operation where the QC log is updated, the updates should be checked into SVN. Multiple users are often completing QC of different files in parallel, which makes checking in updates to the QC log itself important to ensure all records are captured.

logPending

logPending() is a helpful function for identifying files in need of QC. It will return all files that meet the following criteria:

  • assigned with logAssign() but no completed QC
  • modifications made since the last QC was done

Running logPending() on the demo repo will result with the following output:

##                     file                 origin revf headf revo heado reviewer
## 5 script/data-assembly.R script/data-assembly.R    1     5    1     5 michaelm
## 4   script/examp-txt.txt   script/examp-txt.txt    0     1    0     1   anyone
## 6  script/pk/load-spec.R  script/pk/load-spec.R    1     4    1     4 michaelm
##                      time
## 5 2024-06-04 21:04:43 GMT
## 4 2024-06-04 21:04:42 GMT
## 6 2024-06-04 21:04:43 GMT

Looking at script/data-assembly.R, revf = 1 indicates a QC was completed on the file at revision 1. However, headf = 5 indicates edits were made to the file at revision 5, indicating another round of QC is needed.

Looking at script/examp-txt.txt, revf = 0 indicates a QC has never been completed on the file. headf = 1 indicates edits were made to the file at revision 1.

Helpful tools for QCing

When logPending() indicates a file is in need of QC but has previously had a QC completed at an earlier revision, it is helpful to know what changed. This can save QC efforts where only a minor change was made.

diffQced() helps with this as it generates a visual diff in the Viewer pane in R studio. All deleted, added and modified lines of code are shown in the diff.

When diffQced() is supplied a file name, it uses the QC log to identify the revision number when the last QC was completed. The difference is then generated using the version of the file at that revision number versus the current file.

diffQced("script/data-assembly.R")
@@ 1,5 @@
@@ 1,10 @@
 
library(tidyverse)
 
library(tidyverse)
~
>
source(here::here("script", "data-assembly", "da-functions.R"))
 
src_abc <- mrgda::read_src_dir(here::here("data", "source", "STUDY-ABC"))
 
src_abc <- mrgda::read_src_dir(here::here("data", "source", "STUDY-ABC"))
 
derived <- list(sl = list(),tv = list())
 
derived <- list(sl = list(),tv = list())
 
dm_0 <- src_abc$dm %>% filter(ACTARM != "Screen Failure")
 
dm_0 <- src_abc$dm %>% filter(ACTARM != "Screen Failure")
 
derived$sl$dm <- dm_0
 
derived$sl$dm <- dm_0
~
>
pk_0 <- src_abc$pc %>% filter(PCTEST == "TEST OF INTEREST")
~
>
derived$tv$pc <- pk_0
~
>
ex_1 <- src_abc$ex %>% filter(EXTRT == "DRUG OF INTEREST")
~
>
derived$tv$dosing <- ex_1

If you are interested in generating a visual diff with two different files, diffFiles() can be used.

If you are interested in generating a visual diff of a file at specific revision numbers, diffPreviousRevisions() can be used.

QC Summary

renderQCSummary() generates a PDF summary of the QC status of all files checked into the SVN repo.

Some summaries include tables indicating files that are in need of QC (similar to the output of logPending()) and files up to date with QC. Each of these are stratified by the author of the files.

Another helpful summary provided are tables showing all the files checked into SVN, stratified by their QC status. Each contributor to the repo has their own table. The QC status of a given file could be one of the following:

  • Not in QC log: file is checked into SVN, but has not been added to the QC log.
  • In QC log, needs QC: file is in need of QC.
  • QC up to date: file has no QC needs currently.

To see an example of a QC summary document, run the following locally: