Re: Case mapping of sharp s

[Stefan Monnier]

>> > But maybe we're doing something silly somewhere.
>> The Emacs 22 multibyte scheme likely had worse properties for reverse
>> searching.  So maybe something might be simplified nowadays.

> Nope.  The basic nature of the representation and even algorithms are
> the same.  The main difference is that the leading-byte to character
> length map in Mule coding is somewhat arbitrary, while in UTF-8
> there's an algorithm for computing it.  In both cases, the sane
> algorithm is to keep a 256-entry table of corresponding lengths and
> use the octet as an index into that table.

Actually, there is a significant difference when going backward, because
of the special encoding used for eight-bit-graphics in Emacs-22 and the
more regular and less space-efficient encoding used for eight-bit chars
in Emacs-23.


        Stefan