[5] http://www.asahi-net.or.jp/~eb2m-mrt/charsetDetection.html
I was quite surprised you didn't address the role of byte order marks.
BOMs are not limited to XML, neither do they depend on MIME or HTTP. Of
course, they can only be used for UTF encodings.
Thanks for this comment. I think that the use of the BOM or Unicode signature
is an example of charset sniffing based on byte patterns. However, I agree
that
my document was not clear enough. I revised it.
Cheers,
--
MURATA Makoto <murata(_at_)hokkaido(_dot_)email(_dot_)ne(_dot_)jp>