- HTML5 validation of docs [#333]
- bugfix vignette
- increase test suite stability (compatible with recent
desc
release) - document changed behavior:
write_codemeta()
no longer takes installed package name.
- Now uses the existing
codemeta
package on CRAN for all core functionality. - This package provides more advanced metadata detection than the minimalist
codemeta
- Return to CRAN
- Skip on CRAN a test relying on internet connection.
- Bug fix: Fix regression in Windows path handling (#315, @ms609).
- CRAN Solaris issues
write_codemetar()
can now be called from anywhere within a package directory structure. (#305, @mpadge)- Breaking change:
write_codemeta()
writes the JSON file atpath
relative topkg
, not the current directory. (#303, @ThierryO) - Added documentation for changing the default branch (#302, @jonkeane)
- Breaking change: Relatedly, it is no longer possible to use
write_codemeta()
on an installed packages, in that case one would have to usecreate_codemeta()
together withjsonlite::write_json()
. - Use R.Version() instead of R.version to allow mocking in tests.
- Bug fix: now able to parse a README where badges are in a table with non badges links.
- Bug fix:
guess_fileSize()
properly handles.Rbuildignore
(#299, @ThierryO). - Bug fix:
create_codemetar()
handles minimal packages (#298, @ThierryO).
-
The use_git_hook argument of write_codemeta() has been deprecated. Solutions for keeping DESCRIPTION and codemeta.json in sync are available in the docs.
-
drops
crosswalk
, [#288]
-
Docs were improved to make a better case for codemetar.
-
Changes in the way codeRepository is guessed. codemetar can now recognize an URL from GitHub, GitLab, Bitbucket, R-Forge among several URLs in DESCRIPTION, to assign it to codeRepository. If no URL in DESCRIPTION is from any of these providers,
guess_github()
is called. -
Adds documentation of internet needs and verbosity to steps downloading information from the web (#270, @Bisaloo)
-
New argument
write_minimeta
forwrite_codemeta()
indicating whether to also create the file schemaorg.json that corresponds to the metadata Google would validate, to be inserted to a webpage for SEO. It is saved as "schemaorg.json" alongsidepath
(by default, "codemeta.json"). This functionality requires thejsonld
package (listed underSuggests
). -
Updated the GitHub action template to only run on pushes to the master branch and added an explanation of how that works to the readme. (@jonkeane)
-
Fix for detecting rOpenSci review badge (@sckott, #236)
-
Fix extraction of ORCID when composite comment (@billy34, #231)
-
Fix bug in crosswalking (#243)
-
Bug fix: the codeRepository is updated if there's any URL in DESCRIPTION.
-
Bug fix: the README information is now updated by codemeta_readme(). Previously if e.g. a developmentStatus had been set previously, it was never updated.
-
Code cleaning following the book Martin, Robert C. Clean code: a handbook of agile software craftsmanship. Pearson Education, 2009. (@hsonne, #201, #202, #204, #205, #206, #207, #209, #210, #211, #212, #216, #218, #219, #220, #221).
-
Use of re-usable Rmd pieces for the README, intro vignette and man pages to reduce copy-pasting.
- address internet timeout issues
- tidy source code
- update test suite to reflect newly available metadata.
write_codemeta()
andcreate_codemeta()
:use_filesize = FALSE
is now the default and the estimation of the file size does not leave any more unwanted files behind PR #239. Furthermore, the way the file size is calculated changed: Before we used the size of the package built withpkgbuild::build()
, which took rather long. Now the size is calculated based on the source files minus files excluded via
.Rbuildignore
(if such a file exists).write_codemeta()
: the default of argumentuse_git_hook
is nowFALSE
to avoid an unwanted alteration of the user's git environment issue #240.- Package dependency to 'pkgbuild' has been dropped.
write_codemeta()
does not crash anymore if theCITATION
file contains a linecitation(auto = meta)
Issue #238.
-
jsonld
that is used only incrosswalk()
andcodemeta_validate()
is now an optional dependency (Suggests rather than Imports). -
The CodeRepository URL is now cleaned a bit (removing direct link to the README).
-
write_codemeta()
gains a new argumentuse_git_hook
to make the creation of a DESCRIPTION/codemeta.json git pre-commit hook optional. -
create_codemeta()
andwrite_codemeta()
gain a new argumentuse_filesize
to make the building of the package to get its size optional. -
Encoding bug fixed in
extract_badges()
. -
extract_badges()
now usescommonmark
andxml2
instead of only regular expressions. -
pkgbuild
is now used directly instead ofdevtools
. -
give_opinion()
now recognizes lifecycle badges, not only repostatus.org badges.
-
New functions
-
extract_badges for extracting information from all badges in a Markdown file.
-
give_opinion giving opiniated advice about package metadata
-
-
Changes to the create_codemeta output
-
relatedLink field now include provider URL and URL(s) from DESCRIPTION that are not the code repository
-
maintainer is now a list allowing for several maintainers since e.g. the BioConductor a4 package has two maintainers.
-
if more than one CI service among Travis, Appveyor and Circle CI are used and shown via a README badge they're all added to the contIntegration field. URLs from codecov and coveralls badges are also added to the contIntegration field.
-
repo status inferred from the README now 1) is an URL instead of a word 2) recognizes either repostatus.org or Tidyverse lifecycle badges.
-
if present, priority is given to the Repository and BugReports fields of DESCRIPTION for filling the codeRepository and issueTracker fields of codemeta.json (which means working on a fork won't change these).
-
ability to parse all CRAN-allowed MARC roles.
-
if there is a badge for an rOpenSci onboarding review and the review issue is closed, basic review metadata is added to codemeta.json
-
For dependencies, if the provider guessed is CRAN or BioConductor, their canonic CRAN/BioConductor URL is added to codemeta.json as sameAs, unless there's a GitHub repo mentioned for them in Remotes in DESCRIPTION, in which case sameAs is that GitHub repo.
-
CRAN is now correctly translated as "Comprehensive R Archive Network"
-
If codeRepository is guessed to be a GitHub repo (via the URL field of DESCRIPTION or via git remote URL), the repo topics are queried via GitHub API V3 and added to the keywords (in combination with keywords stored in the X-schema.org-keywords field of DESCRIPTION)
-
SystemRequirements are now parsed using https://sysreqs.r-hub.io/, outputting URLs then stored in softwareRequirements
-
-
Help to remind to update codemeta.json regularly: Writing codemeta.json for the first time adds a git pre-commit hook and suggests adding a release question for devtools::release.
-
Internal changes
-
Now uses desc to parse DESCRIPTION files.
-
Package license changed to GPL because of code borrowed from usethis
-
Uses crul instead of httr and uses crul to check some URLs.
-
write_codemeta only uses Rbuildignore and a pre-commit git hook if the function is called from a package folder directly and with the path argument equal to "codemeta.json"
-
The calls to available.packages() for guess_provider now happen inside memoised functions.
-
codemeta_readme function.
-
- Default to DOI-based schema. (previous CN issues now resolved)
- Allow vignettes to gracefully handle network timeout errors that may occur on CRAN's Windows build server.
- CRAN release
- Switch to https://purl.org/ based URIs for the JSON-LD Context file instead of a DOI, due to frequent failure of content negotiation on DataCite servers (#34)
- bugfix UTF-8 characters in CITATION files (#44)
- bugfix to git URLs
- Use
https
on ORCID@id
URIs
- JOSS release
- Post onboarding release
- Added a
NEWS.md
file to track changes to the package.