Re: Weird charset, TWX

From: Tony Harminc (
Date: Fri Sep 26 1997 - 13:43:39 EDT

On 25 Sep 97 at 10:44, Tex Texin wrote:

> This question came to me. The encoding is strange. I suspect that parity
> bits are confusing things and perhaps this is an encoding specific to
> Telex. Anyone have any ideas what this is?

It's absolutely normal ASCII, specified in an odd way.

> The system claims to operation 'under a TTY (TWX) asynchronous dial
> up mode' to support 'other than IBM manufacturers display terminals'.
> It is running on an IBM-3081 with a CICS-BTAM telecommunications handler.
> They also claim they are using an ASCII data stream at 1200 baud,
> using 202 and 103 Bell compatible modems. They also say that the
> HEX representations are MARK parity.
> I don't know if any of the above means anything.

You'll see normal ASCII, though you'll have to ignore the parity
bit. (They claim MARK parity, but it looks more like EVEN to me.)
The reversed notation occurs because of a historical difference of
opinion on which order the bits in a byte should be serialized in -
whether you should visualize shifting the byte onto the wire to the
left or to the right. (I don't remember who does it which way, but
IBM and the ASCII world do it differently.) The mainframe
application program will send EBCDIC bytes (say X'C1' for A) to the
"telecommunications handler", which will have a table that will
translate into reversed ASCII (say X'82'). The UART on the mainframe
end then shifts the bits onto the line as 0 1 0 0 0 0 0 1, which the
receiving UART assembles as X'41'. (Or the other way around, but the
result is the same.)

The only place the weird reversed notation appears is in the
specification of the translate table. Neither you nor the mainframe
application needs to deal with it.

Now why this system is using a protocol that was popular around 1968
is a whole 'nuther question...

Tony Harminc

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:37 EDT