vignettes/datasets.md

In Ironholds/practice: Tools for identifying whether R packages follow common software engineering conventions

Dataset Setup and Usage

Oliver Keyes and David Robinson r Sys.Date()

The practice package provides some datasets, particularly about CRAN, that make it easier to use. For maximum transparency, it is set up to construct these datasets using its own functions.

library(practice)

CRAN Metadata

For example, the CRANmetadata object contains a list of lists, each representing metadata information about a CRAN package. It ships with the practice package.

str(CRANmetadata, max.level = 1, list.len = 6)

## List of 7905
##  $ A3                         :List of 8
##   .. [list output truncated]
##  $ aaMI                       :List of 9
##   .. [list output truncated]
##  $ abc                        :List of 9
##   .. [list output truncated]
##  $ abc.data                   :List of 8
##   .. [list output truncated]
##  $ ABCanalysis                :List of 8
##   .. [list output truncated]
##  $ abcdeFBA                   :List of 8
##   .. [list output truncated]
##   [list output truncated]

str(CRANmetadata[[1]], max.level = 2)

## List of 8
##  $ _id     : chr "A3"
##  $ _rev    : chr "4-a5fc0ad069abe19f1056fb86557ae044"
##  $ name    : chr "A3"
##  $ versions:List of 2
##   ..$ 0.9.1:List of 17
##   ..$ 0.9.2:List of 18
##  $ timeline:List of 2
##   ..$ 0.9.1: chr "2013-02-07T10:00:27+00:00"
##   ..$ 0.9.2: chr "2013-03-26T19:58:40+00:00"
##  $ latest  : chr "0.9.2"
##  $ title   : chr "A3: Accurate, Adaptable, and Accessible Error Metrics for<U+000a>Predictive Models"
##  $ archived: logi FALSE

You can create or update this data yourself, using the update_CRAN_metadata() function:

practice:::update_CRAN_metadata(verbose = TRUE)

Note: as you might expect, this will take several minutes! You can always explore the pre-built CRANmetadata object first and update later at your convenience. If you do update the object, remember to re-install the package via, e.g. devtools::install().

Package sources

The sources of CRAN packages do not come included with the package, since they're ~ 5 GB and counting. However, practice provides a function to download them yourself:

download_packages(get_package_names(), "cran_sources", quiet = FALSE)

This will download the sources of all packages into the cran_sources directory.

Package practices

The CRANpractices dataset contains information on all CRAN packages and whether they follow software engineering practices. Once the package sources are downloaded as above, it can be created with:

practice:::update_CRAN_practices(src_dir = "cran_sources")

(This takes about two minutes). The data looks like:

library(practice)
library(dplyr)

## 
## Attaching package: 'dplyr'
## 
## The following object is masked from 'package:stats':
## 
##     filter
## 
## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

CRANpractices

## Source: local data frame [6,551 x 14]
## 
##        package casing alphanumeric      license  links_to upstream_repo
## 1           A3  Upper         TRUE   GPL (>= 2)  <chr[5]>    None/Other
## 2          abc  Lower         TRUE   GPL (>= 3)  <chr[5]>    None/Other
## 3  ABCanalysis  Mixed         TRUE        GPL-3  <chr[3]>    None/Other
## 4     abcdeFBA  Mixed         TRUE        GPL-2  <chr[7]>    None/Other
## 5  ABCExtremes  Mixed         TRUE        GPL-2  <chr[2]>    None/Other
## 6     ABCoptim  Mixed         TRUE   GPL (>= 3)  <lgl[1]>        GitHub
## 7        ABCp2  Mixed         TRUE        GPL-2  <chr[1]>    None/Other
## 8     abctools  Lower         TRUE   GPL (>= 2)  <chr[6]>    None/Other
## 9          abd  Lower         TRUE        GPL-2 <chr[13]>    None/Other
## 10        abf2  Lower         TRUE Artistic-2.0  <lgl[1]>    None/Other
## ..         ...    ...          ...          ...       ...           ...
## Variables not shown: versioning (lgl), testing (chr), roxygen (lgl),
##   changelog (lgl), vignette_format (chr), vignette_builder (chr),
##   downloads (int), links_from (dbl)

You can use it to answer questions about the frequency of particular practices:

CRANpractices %>%
  count(license) %>%
  arrange(desc(n))

## Source: local data frame [103 x 2]
## 
##         license    n
## 1    GPL (>= 2) 2611
## 2         GPL-2 1433
## 3         GPL-3  702
## 4           GPL  378
## 5    GPL (>= 3)  305
## 6           MIT  226
## 7   GPL-2/GPL-3  150
## 8        LGPL-3   73
## 9  GPL (>= 2.0)   72
## 10 BSD_3_clause   58
## ..          ...  ...

CRANpractices %>%
  count(vignette_format, vignette_builder) %>%
  ungroup() %>%
  arrange(desc(n))

## Source: local data frame [17 x 3]
## 
##       vignette_format   vignette_builder    n
## 1                None               None 5227
## 2               LaTeX             Sweave  898
## 3            Markdown              knitr  240
## 4               LaTeX              knitr  126
## 5                 rsp              R.rsp   21
## 6      LaTeX/Markdown              knitr   10
## 7            Markdown             Sweave   10
## 8               LaTeX          highlight    7
## 9      LaTeX/Markdown       utils, knitr    2
## 10 LaTeX/Markdown/rsp       knitr, R.rsp    2
## 11                rst              knitr    2
## 12              LaTeX elliptic, emulator    1
## 13              LaTeX              noweb    1
## 14              LaTeX              utils    1
## 15          LaTeX/rst             Sweave    1
## 16           Markdown  knitr, RefManageR    1
## 17       Markdown/rsp       knitr, R.rsp    1

Ironholds/practice documentation built on May 7, 2019, 6:41 a.m.