[Info-vax] character set translation for language accents

[P. Sture]

In article 
<02218112-3ac8-4bb5-b406-50e559226c93 at r34g2000vba.googlegroups.com>,
 jcwoman1963 at hotmail.com wrote:

> Here is more info.  I'm using a tcpip socket for my communications
> path.  Luckily, on the VMS side my device driver and application are
> written in Macro, so I'm able to debug and troubleshoot them to my
> heart's content.  I have a "trace" function written into the driver
> that simply copies the data it receives from the socket to my screen,
> along with a hex representation of the same data.  So I found that
> while the data is being displayed as (null, I assume), the hex appears
> to be correct.
> 
> For example, when I receive this data:
> 
> Vous répondez à une personne....
> 
> my system displays it as this:
> 
> Vous rpondez  une personne....
> 
> but the hex translation is this:
> 
> 56 6f 75 73 20 72 e9 70 6f 6e 64 65 7a 20 e0 20 75 6e 65 20 70
> 65 .....
> 
> Since it sees that first  é    as hex e9, does that mean it's not
> truncating the high order bit, or still is?  If I'm not being
> truncated, I can easily add my own translation table to make the hex
> e9 = a plain old e.  It's not ideal for language, but I think this
> customer is used to our system not having accents, so it would
> probably be more acceptable than just dropping letters out of words.

Yes, e9 does have the high order bit set. Try this:

$ write sys$output %xe9
233

alternatively, do

$ x=%xe9
$ show symbol x
 X = 233   Hex = 000000E9  Octal = 00000000351

-- 
Paul Sture