Julian Reschke wrote:
those who still use operating systems with no builtin support
for UTF-8 are knowledgeable enough to run a text file - in
particular if it has a UTF BOM - through GNU recode.
A mandatory BOM for those I-Ds / RFCs actually using UTF-8 is
an idea for Paul's next draft. Just for fun I test it here.
I don't know "GNU recode" so far - my UTF-8 script would only
recode characters with a 1:1 correspondence in windows-1252.
So maybe Paul would also need "fully normalized" as defined
in http://www.w3.org/TR/charmod-norm/ (that's not yet ready).
Ietf mailing list