Update.

Update.

	* posix/regex_internal.h (re_token_type_t): Remove unused ALT,
	END_OF_RE_TOKEN_T and SUBEXP.  Reorder values.  Add OP_UTF8_PERIOD
	and EPSILON_BIT.
	(IS_EPSILON_NODE): Just test if EPSILON_BIT is set.
	(ACCEPT_MB_NODE): Return 1 for OP_UTF8_PERIOD as well.
	* posix/regex_internal.c (create_ci_newstate, create_cd_newstate):
	Handle OP_UTF8_PERIOD.
	(re_string_reconstruct): Set valid_len for single byte char searching
	with no translation and case sensitivity.
	* posix/regcomp.c (re_compile_fastmap_iter, calc_first): Handle
	OP_UTF8_PERIOD.
	(re_compile_internal): Don't call optimize_utf8 if preg->translate
	!= NULL.
	(optimize_utf8): Remove BACK_SLASH case.
	Transform OP_PERIOD into OP_UTF8_PERIOD if the searching can be
	optimized.
	(parse_bracket_exp): Don't create SIMPLE_BRACKET if it doesn't have
	any bits set and COMPLEX_BRACKET is used.
	* posix/regexec.c (transit_state_mb): Fix comment typo.
	(group_nodes_into_DFAstates, check_node_accept): Handle
	OP_UTF8_PERIOD.
	(check_node_accept_bytes): Likewise.  Reorder slightly so that
	re_string_char_size_at and re_string_elem_size_at are called
	only when needed.
	* posix/bug-regex20.c (BRE, ERE): Define.
	(tests): Use them to make lines shorter.  Expect . to be
	optimized.  Add lots of new tests.
	(main): Run (ATM just case sensitive) test with backwards searching
	as well.

2003-11-18  Jakub Jelinek  <jakub@redhat.com>