Dfc optimization

[Ben Wing]

> I assume that mswindows-multibyte has similar facilities.

No.  It's just a catchall for any code page, which can pretty much do
whatever it wants.  It's similar in spirit to the "encoding" setting of a
locale.  Depending on the locale, you have a code page, and the code page
specifies a way to encode sequences of characters into strings.  *Most* code
pages use either SBCS or SBCS/DBCS non-modal encodings.  But I think there
is a UTF8 code page and it's even vaguely possible there is a JIS code page
[I'm not sure of this; it seems that internally the mapping between code
page value strings and Unicode characters is handled through table lookup --
this means that a modal encoding is out; but any EUC variety would be
allowed.]

So basically you have to look at the code page and make sense based on it.