🚀 go-pugleaf

I know we covered this before and I was assured that each SJIS character
has a Unicode equivalent, but here's a wrinkle:

Hitachi tells me that on their workstation (which supports SJIS and J/EUC)
there is a one-byte '1' and a two-byte '1' (as just one example). I'm
not sure if this is standard SJIS or not, but I believe it comes from
the one-byte ASCII '1' added to the two-byte JIS '1'.

Anyway, since our application will receive identifiers such as filenames
from other applications, and these may contain one-byte '1's and/or
two-byte '1's, we MUST pass them back as we received them.  Converting
them into an internal Unicode character set and then back actually
will result in data-loss.

Am I correct that this is a true gotcha, or does Unicode provide
some solution for this?

[Another wrinkle is an area set-aside in SJIS for user-defined
characters, such as company logos.  We must store and return
these characters as well.  Does this map into Unicode?]

Thanks,
Steve Billings
srb@world.std.com