2.3.0
  - Support for spellchecking on Quarto files (@olivroy, #77)
  - You can now specify a global wordlist file via the SPELLING_WORDLIST envvar

2.2.1
  - Fix for commonmark 1.7.0 change for auto-linked markdown URLs

2.2
  - WORDLIST is now sorted with a locale-independent method, which avoids large
    diffs in version control due to the fact that developpers use different
    locale, with different lexicographic ordering rules (#48, @bisaloo)
  - spell_check_package() now loads Rd macros (#42)

2.1
  - Pre-filter script/style/img tags when checking html files because the huge
    embedded binary blogs produced by rmarkdown slow down the hunspell parser.
  - Treat input files in spell_check_files() as UTF-8 on all platforms
  - Fix a sorting bug in spell_check_files()

2.0
  - spell_check_package() now also checks README.md and NEWS.md in the package root
  - Enforce latest hunspell and libhunspell, which include updated dictionaries
  - Treat all input as UTF-8. Fixes some false positives on Windows
  - Ignore yaml front matter in markdown except for 'title', 'subtitle', and 'description'
  - Markdown: filter words that contain an '@' symbol (citation key or email address)
  - Properly parse authors@R field for ignore list (issue #2)
  - Use tools::file_ext instead of knitr:::file_ext

1.2
  - Internally normalize all case of lang strings to lower_UPPER e.g en_US
  - Only run automatic check when 'spelling' is available and NOT_CRAN is set

1.1
  - Breaking: Package spell-checker now uses language from DESCRIPTION
  - Require hunspell 2.9 dependency (better parsing and dicationaries)
  - Change default lang to 'en_US'

1.0
  - Initial release