Re: [Gnu-arch-users] Re: How does arch/tla handle encodings?

[Tom Lord]

    > From: Aaron Bentley <address@hidden>

    > The mantra seems to be "perform operations on the string, not
    > the codepoint".  Even case conversions seem to work best on
    > strings, not codepoints.


Not just "seem to" --- are "defined to".   

Except when they aren't.

Nearly all such questions I've discovered so far are well addressed by
the Unicode consortium publications.  It ain't light reading.  It
ain't simple or trivially clean.  It don't look wrong to me.  (Turns
out letters and strings aren't all that simple or trivially clean, in
general --- with ASCII, we got a lucky simplified special case.
Unicode at least places an upper bound on the complexity and gives a
good start on using it in practice.)

-t