In transliterations:
characters are VERY literal, except for - not at the start or end
- of the string, which indicates a range. If the range is in bytes,
+ of the string, which indicates a range. However some backslash sequences
+ are recognized: \r, \n, and the like
+ \007 \o{}, \x{}, \N{}
+ If all elements in the transliteration are below 256,
scan_const expands the range to the full set of intermediate
characters. If the range is in utf8, the hyphen is replaced with
a certain range mark which will be handled by pmtrans() in op.c.
In double-quoted strings:
backslashes:
- double-quoted style: \r and \n
- constants: \x31, etc.
+ all those recognized in transliterations
deprecated backrefs: \1 (in substitution replacements)
case and quoting: \U \Q \E
stops on @ and $