ANN: pandoc 2.0

[John MacFarlane <jgm-TVLZxgkOlNX2fBVCVOL8/A@public.gmane.org>]

I'm delighted to announce the release of pandoc 2.0.
Binary packages and a changelog can be found at
<https://github.com/jgm/pandoc/releases/tag/2.0>.  The source
code and API documentation is on Hackage:
<https://hackage.haskell.org/package/pandoc>.

This is a huge release:  describing all the changes takes over
11,000 words in the changelog.  All users should look at the
"new features" and "behavior changes" sections of the
changelog.  Those who use pandoc as a library should also look
at the "API changes" section.  And anyone who works with a
particular format should look at the "bug fixes" section
for relevant to that format.

Some highlights:

- New output formats:  ms (groff ms), jats (JATS XML), gfm (a
  version of GitHub-Flavored Markdown that uses the same
  parser as GitHub for maximum accuracy), muse (Emacs Muse).

- New input formats:  gfm, muse, tikiwiki (TikiWiki),
  vimwiki (Vimwiki), creole (Creole 1.0).

- A plain-text syntax for Divs (arbitrary block containers).

- A syntax for passing through raw content in any format.

- Much improved support for LaTeX input, especially
  in handling macros and included files.

- PDFs can now be produced via pdfroff, prince, and weasyprint,
  in addition to latex, xelatex, lualatex, context, and
  wkhtmltopdf.

- A new way of writing pandoc filters in lua, using the lua
  interpreter that is built into pandoc.  Lua filters are
  generally much faster than JSON filters (since we avoid
  the expense of converting to and from JSON), and they have
  the advantage that they do not require any software besides
  pandoc itself to be installed.

- Better error handling, warnings and informational messages.

- New command-line options:

  --eol (specifies line endings),
  --log (prints JSON representation of info and warning messages),
  --request-header (allows specifying a header to be used when
    pandoc fetches external resources),
  --lua-filter (for running lua filters, see below),
  --epub-subdirectory (for changing the directory used in epub
    containers),
  --resource-path (for setting the search path for images and
    other resources),
  --abbreviations (for specifying a custom abbreviations
    file so that the Markdown parser can be sensitive to
    abbreviations),
  --syntax-definition (allowing XML syntax highlighting definitions
    to be loaded dynamically).

- Changed command-line options:

  --reference-doc replaces --reference-docx and --reference-odt
  --smart removed (instead use +smart on reader and/or writer)
  --normalize removed (normalization is automatic)
  --latex-engine removed (instead use --pdf-engine)
  --parse-raw removed (instead use -f latex+raw_tex or -f html+raw_html)
  --epub-stylesheet removed (instead use --css)
  --mathml no longer takes an argument

The focus of this release was a major architectural change
that will be largely invisible to users of the pandoc program,
but that is responsible for many of the improvements users
will notice.  Previously, most of the pandoc readers and
writers were "pure" functions: they converted between strings
and a Pandoc data structure, without being able to perform IO
operations.  This purity has significant advantages in many
contexts, but it also makes it difficult to do things like
process included files.  We have changed the types of all the
readers and writers so that users can now select whether
they will have access to IO.

With each release, pandoc becomes more a team effort, with more
contributors committing high-quality code.  Special thanks are
due to Jesse Rosenthal, who designed and initiated the large
architectural change described above; Albert Krewinkel,
who added the lua filters and improved the Org reader, as well
as helping with the API reorganization; Alexander Krotov,
who contributed the Muse reader and writer and improved the
FB2 writer; Andrew Dunning, Vaclav Haisman, Xavier Olive, and
Thomas Hodgson, who improved the LaTeX, ConTeXt, and beamer
templates; Kolen Cheung, who improved documentation and
infrastructure (pandoc-nightly); Marc Schreiber, who improved
the LaTeX reader; Mauro Bieg, who improved the LaTeX reader and
image size code, and added PDF generation via weasyprint and
prince; Sascha Wilde, who added the Creole reader; Yuchen Pei,
who added the Vimwiki reader; hftf, who brought consistency to
the readers' handling of underlining; and rlpowell, who added the
TikiWiki reader.  Many others contributed code, bug reports, or
suggestions.  Without this large and cooperative community, pandoc
would be a far less capable tool.

Happy converting!