[MLton] Unicode / WideChar

[Henry Cejtin]

The problem with number conversion and the like (lexicographic ordering of
strings, etc.) being locale dependent is really REALLY severe.  The point is
that if I run a program, often (almost always?) the input and output come from
other programs and are NOT primarily intended for human consumption/production.
In that case, you really do NOT want locales entering in to things at all.

I don't know how to combine that with the notion of adapting things like
locale-specific formats.  In the end, I just gave up and set my LANG to C
as a hack to avoid the problem.