Version 0.2.4:
    Bug fixes:
    - Compile with Large File Support if such support is available on the
      platform.

Version 0.2.3:
    Bug fixes:
    - Fix memory leak in t3_highlight_detect.

Version 0.2.2:
    New features:
    - Added LaTeX output option for t3highlight program.

    Bug fixes:
    - Fix missing casts to unsigned char in UTF-8 validity check.

Version 0.2.1:
    New features:
    - A new API for autodetection of languages using either the vi(m) modelines
      and Emacs major mode comments, or the interpreter for scripts.
    - Switch to XDG Base Directory Specification locations for per-user
      configuration files.

Version 0.2.0:
    New features:
    - Added dynamic 'end' patterns, which allows the 'end' pattern to be
      dependent on the text matched by the 'start' pattern.
    - Allowed 'end' and other patterns to specify how many states to exit. This
      also allows multiple 'end'-like patterns and error patterns.
    - Added the possibility to push extra states on the stack on matching a
      'start' pattern. This allows more state based handling of the input.
    - Added perl language definition.
    - Added HTML language definition.
    - Added CSS language definition.
    - Added MathML language definition.
    - Added SVG language definition.
    - Updated C language definition to C11 standard.
    - Improved the handling of here-docs in the shell language.
    - Implemented internal UTF-8 validity check, such that pcre_exec can be
      called with PCRE_NO_UTF8_CHECK. This increases performance.
    - Changed the matching routine to use anchored matching, to speed up the
      matching process. As a result, the \G assertion matches everywhere, and
      is therefore rendered useless. This results in a large (>10 times)
      speed-up.
    - Added t3highlight executable, which allows using the highlighter from the
      command line. Different output styles can be used, with styles for
      terminal output (esc) and HTML included.
    - Added verbose error reporting.

    Bug fixes:
    - Don't use \b to determine word boundaries when - is also a word character.

Version 0.1.3:
    Bug fixes:
    - Fix memory leak in loading available names.

Version 0.1.2:
    New features:
    - Added python and python3 language definitions.

    Bug fixes:
    - Improved floating point constant matching for Java and C/C++.
    - Use longest match instead of first match as tie breaker.

Version 0.1.1:
    Initial release
