NEWS

1.3.0 (2015-09-08):
 - Added python 3 support.
 - Added an example of parsing ISO 8601.
 - Support rule arguments in parser protocols.

1.2.0 (2013-08-01):
 - Added 'makeProtocol' for parsing stream protocols using Parsley
   grammars and Twisted, with examples, docs, and a tutorial.
 - Added 'stack' as support for 'makeProtocol'.
 - Sped up parsing; started compiling python expressons only once.
   (thanks to eevee!)
 - Allowed tuple unpacking in assignment in rules. For example,
   `rule1 = rule2:(a, b) -> a + b`. (thanks to introom!)
 - Added a trace hook and an example of a trace visualizer using it.
 - Fixed an example in the README. (thanks to ilmarga!)
 - Added support for custom labels in grammars. For example,
   `rule1 = rule2 ^(foo)` will make Parsley consider 'foo' the name for
   'rule2' in error messages. (thanks to miebster!)

1.1.0 (2013-03-03):
 - The 'spaces' rule has been provided under the alias 'ws'; 'spaces'
   is now the deprecated name.
 - Single quotes can now be used around multiple characters to match
   them consecutively. For example, 'abc' is equivalent 'a' 'b' 'c'.
 - The 'token' rule is thus deprecated as well, since "token('foo')"
   is more typing than "ws 'foo'".
 - Hex escapes can now be used in literals for matching bytes.
 - 'makeGrammar' now takes an 'unwrap' argument, to get a grammar
   class suitable for subclassing.
 - 'makeGrammar' also takes an 'extends' argument that can be either a
   regular python class or a wrapped Parsley grammar, for simple
   subclassing.
 - 'repeat' expressions now work for 0 repetitions properly.
 - ometa.runtime.TIMING flag added for enabling debug output of
   grammar load times.
 - OMeta version 1 parser moved to ometa.compat.
 - Twine removed.
 - Various bugfixes around error reporting.
 - Parsley no longer loads its internal grammars at import time. The
   'stage' script in bin/ is used to regenerate the compiled versions
   after modification.
 - Experimental tree transformer grammars, with special syntax for
   destructuring terms and creating string templates, have been added.

1.0.0 (2012-10-01):
 - PyMeta is now Parsley.
 - A new public API has been added in the 'parsley' module.
 - Grammars now use the OMeta 2 syntax by default. Compatibility is
   available via ometa.grammar.OMeta1.
 - A generic AST-handling library has been added, 'terml'.
 - Grammars now parse to term trees instead of nested lists. The code
   generator has accordingly been greatly simplified.
 - A grammar interpreter has been added, which parses by evaluating
   the grammar term tree directly instead of generating a class.
 - A push parser has been added which accepts data in arbitrary
   chunks, maintaining state while waiting for more input.

0.4.0 (2010-05-15):
 - Builders now work from a grammar AST, rather than being driven by
   the grammar parser directly.
 - All parse operations now carry error information, including what
   was expected to be parsed and where.
  * Consequently, ParseError now has a 'formatError' method that takes
    the input string and returns a human-readable description the
    parse failure.
 - New example: TermL parser, in examples/terml.
 - New script, bin/generate_parser. Takes a file containing a PyMeta
   grammar as input and writes a Python module file.

 - A couple bugfixes:
 #248643 subclassed grammars don't inherit base namespace
 #564135 makeGrammar error handling for nonsensical grammars is broken


0.3.0 (2008-07-12):
 - Input stream now implemented like the Javascript version, with immutable
   position objects.

0.2.0 (2008-04-17):
 - Grammars generate Python source instead of ASTs.
 - OMeta now has a "metagrammarClass" attribute, to ease extension of
   the metagrammar.

0.1.1 (2008-03-27):
 - Fixed a bug in 'super'.

0.1 (2008-03-26):
 - Initial release.