GitHub vs GitLab for most recent 1000 R package releases

github_vs_gitlab.R

# Measure usage of GitHub vs GitLab in R packages. This likely undercounts both GitHub and GitLab, 
# becuase not every package puts a link to the development repository in their URL or BugReports fields.

# remotes::install_github("r-hub/crandb")

last_1000 <- crandb::events(archivals = FALSE, limit = 1000)

has_url <- function(x, url) {
  any(grepl(url, c(tolower(x$package$URL), tolower(x$package$BugReports)), fixed = TRUE))
}

get_pkg_info <- function(x) {
  list(package = x$package$Package,
    maintainer = x$package$Maintainer %||% NA_character_,
    version = x$package$Version,
    date = x$date
  )
}

library(purrr)

all_releases <- map_dfr(last_1000, get_pkg_info)
length(unique(all_releases$maintainer))
#> [1] 735

is_github <- map_lgl(last_1000, has_url, url = "github")
github_releases <- map_dfr(last_1000[is_github], get_pkg_info)

sum(is_github)
#> [1] 560
length(unique(github_releases$maintainer))
#> [1] 401

is_gitlab <- map_lgl(last_1000, has_url, url = "gitlab")
gitlab_releases <- map_dfr(last_1000[is_gitlab], get_pkg_info)

sum(is_gitlab)
#> [1] 15
length(unique(gitlab_releases$maintainer))
#> [1] 10

# Created on 2019-08-29 by the [reprex package](https://reprex.tidyverse.org) (v0.3.0)