8.2.1

8.2.1

• FIX: Reject compiled replace templates from another search pattern in compiled search objects,
  matching module-level substitutions and preventing incorrect capture references.
• FIX: bre: Fix a - after a POSIX class in a character set sometimes forming a range ([[:blank:]-z]).
• FIX: bre: Fix an escaped - before a POSIX or Unicode class in a character set being turned into a
  range operator ([\-\p{N}]).
• FIX: bre: Fix ^ being treated as a negation when it follows properties in a character set that match
  nothing, such as \p{Han} in a byte string ([\p{Han}^a]).
• FIX: bre: Fix ranges in byte string character sets with a named character beyond the byte range
  ([a-\N{CJK UNIFIED IDEOGRAPH-4E00}]). They now match Regex: a range that ends beyond the byte range includes
  everything up to \xff, and a range that starts beyond it matches nothing, or is an error if it is reversed.
• FIX: bre: Fix a literal ^ after the negating ^ in a character set ([^^]) causing an unterminated
  character set error.
• FIX: bregex: Fix finding the end of character sets, which could cause false errors or wrongly handle what
  follows a set (such as \R, groups, or verbose comments). Character sets now follow Regex's own rules:
  • POSIX classes are recognized in byte strings, and with qualified names or values ([[:script=latin:]]).
  • In V1, a ] directly after a set operator is a literal ([\w--]]).
  • A literal ^ after the negating ^ ([^^]) no longer causes an unterminated character set error.
• FIX: bregex: POSIX classes in character sets are no longer rewritten before being passed to Regex.