additional conversions are needed at every step. Some problems
remain, but they all seem to be related to the combination of
the extra mapping just described and case-insensitive matching:
-for example, "\x{131}" (LATIN SMALL LETTER DOTLESS I) does not
-match "I" case-insensitively, as it should under Unicode.
+for example, C<< /[\x{131}]/ >> (LATIN SMALL LETTER DOTLESS I)
+does not match "I" case-insensitively, as it should under Unicode.
(The match succeeds in ASCII-derived platforms.)
In any case, the Unicode support on EBCDIC platforms is better than