2014-03-17  Kaz Kylheku  <kaz@kylheku.com>

	* awkreg.awk (eat_bracket_exp): Capture the endpoint characters of a
	range into the variables x and y, and use these to assert that
	the lower endpoint is not lexicographically above the upper endpoint.

2014-03-17  Kaz Kylheku  <kaz@kylheku.com>

	* awkreg.awk (driver): fix references to nonexistent variable ir;
	should be ok.

2014-03-17  Kaz Kylheku  <kaz@kylheku.com>

	* awkreg.awk (eat_bracket_exp, eat_regex): Remove unnecessary
	statement-terminating semicolons.

2014-03-17  Kaz Kylheku  <kaz@kylheku.com>

	* awkreg-grammar.txt: New file. Taken from my original Usenet
	posting in comp.lang.awk.

2014-03-17  Kaz Kylheku  <kaz@kylheku.com>

	Fix in {m,n} syntax.

	The issue is that the parser partially consumes broken {m,n}
	syntax. The eat_rep_notation function must backtrack fully,
	and its caller must detect it has done so.

	Improved driver code.

	* awkreg.awk (dbg): New function, handy for debugging.
	Prints its argument, and then returns it.
	(match_and_eat_else): New function. Similar to match_and_eat,
	except that if there si no match, it returns the new third
	argument, rather than the rejected string itself.
	This simplifies the expression of some backtracking logic.
	(eat_rep_notation): If any of the notation doesn't match,
	then backtrack all the way and return the original input unconsumed.
	For this, a new local variable o ("original") retains the input.
	(eat_factor): When matching a parenthesized regex, if matching closing
	parens after the regex fails, then backtrack all the way,
	returning the whole factor unconsumed.
	(regex_check): New function: contains most of the logic of
	the original is_regex. Basically it is like eat_regex, but with support
	for ^ and $.
	(is_regex): Basically now just reports whether regex_check fully
	consumes the string or not.
	(driver): The driver action is improved, showing the non-matching
	suffix of the regex.