Adding authors with long names or other fields (comment, typically)
works well now (#91).

get_deps() now removes unneeded whitespace from version requirements
(#84).

normalize() now does not drop Authors@R on non-UTF-8 systems
when it has non-ASCII characters (#80).

has_dep() now works well with dependencies listed multiple times
(#97, @richfitz).

Add coerce_authors_at_r() method to convert Author to
Authors@R (#44, @muschellij2).

add_author() and similar functions now allow a character vector of
multiple roles (@niceume, #89).

desc_set_deps() now inserts new packages in (case-insensitive)
alphabetical order, if the existing packages are already in alphabetical
order.

New add_author_gh() method and desc_add_author_gh() function to add
an author using the information available from GitHub V3 API. This method
and function depend on gh and are limited when the GitHub user full
name is incomplete or not well parsed by as.person() and when their
email address isn't available (@maelle, #71).

When using desc_normalize() the package dependencies are now
alphabetically sorted (#66, @llrs).

New add_orcid() method and desc_add_orcid() functions make it
possible to add ORCID IDs to authors directly instead of via the
comment argument (@maelle, #70).

All functions and methods managing authors (add_me, add_author(),
del_author(), add_role(), del_role(), change_maintainer(),
search_for_author(), add_me(), etc.) gain an orcid argument
(@maelle, #70).

In person() within the Authors@R field, comment can now be a
named character vector (@maelle, #69; @gvegayon, #65).

When using desc(text=) parameter, set
textConnection(encoding = "bytes") to handle cases when the input
text is in a different marked encoding than the default encoding, such
as UTF-8 input on Windows.