Revision history for Perl extension MarpaX::Grammar::Parser.

1.08  Sat Oct 25 17:20:00 2014
	- Bump pre-req version of Tree::DAG_Node to 1.23, because of the new output format generated by
		tree2string().

1.07  Wed Oct 22 16:13:00 2014
	- Bug fix: Clean up search for first rule (in case there is no :start rule).
	- Clean up visual breaks, which should have all be 50 chars long.

1.06  Wed Oct 22 13:37:00 2014
	- Oops. Forgot to run homegrown validate.build.make.pl to cross-check pre-reqs.
		Added List::AllUtils and Types::Standard.

1.05  Sun Oct 19 17:45:00 2014
	- Update BNF from Marpa (share/metag.bnf).
	- Update BNF from MarpaX::Languages::C::AST (share/c.ast.bnf).
	- Update BNF from MarpaX::Demo::StringParser.
	- Add a feature to convert the cooked tree into a hashref. See output_hashref in the docs.
		Sample output is in share/*.hashref.
	- Update the code to correspond to the tree of grammar generated by Marpa, which has changed
		slightly since the last version of this module.
	- Reformat this file so line lengths are 100 chars max.
	- Update the docs generally.
	- Ship share/metag.cooked.tree, share/metag.hashref and share/metag.raw.tree.
	- Ship share/stringparser.hashref.

1.04  Mon Oct  6 15:32:00 2014
	- Bump the required version of Marpa::R2 from V 2.066000 to V 2.096000. This is necessary since
		the forgiving and latm tokens were introduced after V 2.066000, and I was starting to get
		CPAN tester failures.

1.03  Fri Sep 19 11:40:00 2014
	- Handle new Marpa output (from when it dumps its own grammar).
	- Update Build.PL and Makefile.PL about this module being on github.
	- Re-generate share/*.[cooked,raw].*.

1.02  Tue Sep  9 09:57:00 2014
	- Remove redundant Algorithm::Diff from t/runner.t.
	- Switch from Perl6::Slurp to File::Slurp.
	- Update copy of Marpa's metag.bnf.
	- Add share/json.3.bnf.
	- Re-generate share/*.[cooked,raw].*.

1.01  Fri Aug 16 14:53:00 2013
	- Explicitly test for the files marpa_bnf_file and user_bnf_file. Die if missing.
	- Patch Data::TreeDumper::Renderer::Marpa. The first 2 daughters of each node are the start and
		end offsets of the token within the input stream. Their attribute type should be 'Marpa',
		not 'Grammar'.
	- Remove the 'level' attribute from the nodes in the raw tree. It didn't offer any value.
	- Add methods clean_name(), compress_tree(), compress_branch(), cooked_tree() and
		cooked_tree_file().
	- Change the no_attributes option to bind_attributes. This reverses the sense of the option it's
		true, but at least makes it positive logic (as it should have been all along). This was
		bugging me, both here and (potentially) in the next module in the chain,
		Marpa::Grammar::GraphViz2.
	- Expand the docs to cover all the features.

1.00  Tue Aug  6 15:47:00 2013
	  - Original version