FWIW, the change sounds all good to me.  The O(1) is the most important
property of a string, the 8/16/32 gives us that and space savings too,
going all Unicode at the heart of it is the only sensible thing to do
(anything else leads into combinatorial explosion and instant insanity),
"encodings" are just serializations and belong at the earliest to I/O,
and a "language" attribute doesn't belong in a string.  (It may belong
to a string in a higher abstraction level, such as a database, but not
here.)  Good work.

Reply via email to