>One thing MS _really_ got wrong is when you want to have both multi-byte
and
>single-byte support at the same time. It's hell. Rather follow a clean
model
>where the OS offers functions to manipulate both kinds of data depending on
>what you want, like this exists on the Mac OS. Linking against different
>libraries to produce builds that support multibyte is a very bad idea,
IMHO.
i think you are saying "multi-byte" when you mean "unicode". "multi-byte"
and "DBCS (double-byte character set) are specifics term that means "one or
two physical bytes to represent one logical character, depending on the
character you want to represent, but the underlying data type is still
sizeof(Char) == 1". SBCS == "single-byte character set", but SBCS and DBCS
are just different perspectives on the same underlying data type. going
DBCS requires very little effort or overhead.
i think you are refering to the fact that NT has full Unicode support, but
Win9x has essentially zero unicode support, thus making it a very very large
headache to write an app that internally uses unicode as much as possible
but still runs correctly on both Win9x and NT. writing an app that
literally supports both SBCS and DBCS is extremely easy, with almost zero
overhead, because of the definition of how DBCS works.
