3.0.2
  - Find paths to dictionaries of RStudio 1.3.x and 1.4.x (@GegznaV, #42)

3.0.1
  - Fix for BDR such that vignette can be built without 'pdftools' installed

3.0
  - Update bundled libhunspell to v1.7. This seems like a pretty major update, see
    upstream release notes: https://github.com/hunspell/hunspell/blob/master/NEWS
  - Add rstudio directory to the search path to use dictionaries installed via the
    RStudio IDE (Global Options -> Spelling -> Main Dictionary Language -> Install)
  - Updated bundled en_US and en_GB dictionary to Open Office release 2018-11.01:
    https://extensions.libreoffice.org/extensions/english-dictionaries/2018-11.01
  - Workaround for adding custom words containing 'right single quotation mark',
    these now always get converted to regular apostrophe.

2.9
  - BREAKING: Bundled en_US / en_GB dictionaries now support apostrophe
  - Parses now uses the dictionary WORDCHARS for tokenizing text
  - Removed numbers from en_US / en_GB WORDCHARS to prevent false positives
  - Fix regression inroduced in the caching fix in 2.8
  - Refactored internal dictionary caching
  - Add 'AUTHORS' file and update DESCRIPTION as requested by CRAN

2.8
  - Fix caching bug for 'ignore' argument in hunspell()
  - Rename class 'dictionary' to 'hunspell_dictionary' to avoid collisions
  - Remove setwd() from examples as requested by CRAN

2.7
  - Update bundled en_US / en_GB dictionaries from libreoffice extensions
  - Use Rcpp symbol registration / visibility
  - Properly pass down missing values
  - Added workaround for issue #29 (case sensitivity in custom wordlist)

2.6
  - Compile libhunspell with '__attribute__((__visibility__("hidden")))' to solve
    a symbol conflict in rstudio (which also has libhunspell)

2.5
  - Add parameter 'add_words' to dictionary()

2.4
  - Update libhunspell to upstream v1.6.1
  - Update maintainer info
  - Add mandatory symbol registration

2.3
  - Update libhunspell to upstream v1.5.4
  - Change intro.rmd vignette to clean up downloaded files

2.2
  - Tweak code to make it build on old compilers (CentOS6 / gcc 4.4.7)

2.1
  - Update upstream to a6d32ee
  - Rebuild vignettes to fix CMD check timestamp warning

2.0
  - Added a beautiful intro vignette
  - Dictionaries are now their own class and get cached automatically via memoise
  - Make sure UTF-8 return values are marked properly. Fixes #16
  - Update libhunspell to upstream 4b43843

1.4.3
  - Fix UBSAN bug
  - Remove unused 'config.h' file (see upstream 2ccf840)

1.4.2
  - Switch to R's iconv wrapper which is more portable (thnx BDR)

1.4.1
  - Change license to cover libhunspell (per CRAN request).

1.4
  - Switch to bundled libhunspell because their API keeps breaking
  - Include libhunspell 1.5-pre (b13e62a)
  - Add parsers for HTML/XML formats

1.2
  - (Breaking) Rename 'hunspell_find' to 'hunspell'
  - Add support for other dictionaries
  - Use iconv() to convert encoding before checking
  - Use the 'en_stats' dict as default ignore list

1.1
  - Switch to hunspell parsers (replaced 'delim' with 'format' parameter)

1.0
  - Initial CRAN release
