Bing Fu wrote about using FTP to transfer a file from MPE to Windows
NT and how his French characters appeared incorrect after the
transfer.  But using Reflection worked.

I don't have a specific answer to the question, but I can give some
background on the problem.  Characters on MPE (and HP-UX) are stored
using the HP Roman-8 character-set.  Characters on Windows NT are
generally stored as UNICODE (or the MS Windows character set).

Characters in both character sets are represented as a digit number.
Every number corresponds to an individual letter and visa-versa (at
least as far as Latin-ISO-1 characters are concerned).  For the ASCII
characters (letters, numbers, and North American special symbols), the
numeric representation is the same in both HP's Roman-8 character set,
UNICODE, and the MS Windows character set.

But for Latin-ISO-1 characters with a numeric value greater than 127,
there are differences between these character sets.  When you transfer
a file with FTP, the numeric values are correctly transferred to
Windows NT.  But on MPE lower-e-accent is represented by the numeric
value 197, but in Windows NT the value 197 is a capital-A with a
circle above it.

So why does the transfer work with Reflection?  The reason is that
Reflection is aware of these character differences and does a good job
of doing the necessary translations as part of the file transfer.

My guess is that you won't be able to use FTP to transfer the file and
get character translation.  You could use FTP, if you were willing to
write either:

1.  A program on MPE that pre-translated the file before transferring
    it with FTP, or

2.  A program on Windows NT that post-transated the file after
    transferring it with FTP.

For more information about character sets see our SMUG Book entry:

     http://www.robelle.com/smugbook/char.html

Cheers,

David    <[log in to unmask]>