> Let's not forget that Kermit is a protocol that was designed for maximum
> portability. It allows machines of various bits, even 5 bits in the case of
> Baudot to succesfully transfer binary data. ASCII in 7/8 bit mode also has
> this capability, although not always used correctly in 7-bit stripping. It
> is called the Shift In Shift OUT sequence. Essentially if all software would
> QUOTE 8-bit, when converting to 7-bit and back instead of toosing away that
> bit then one to one correspodence for 8-bit streams would be handled the way
> God intended thhem to be handled. The sender of a message should quote the
> message in 7-bit format. This would make most files only slightly larger
> than now.:) and make the incentive to Correctly xlat 7-bits to 8-bits all
> the more urgent. If bit 8 isn't 0 then it is intentional and should be left
> alone. This all hardware/software Compatable 100%, and provide and incentive
> to LEARN from efforts like NAPLPS, and videotext.
>
I don't think anybody learns from anything any more. After all, the current
desktop revolution seems to have tossed out the entire corpus of computer
science pre-circa-1990, back in the days when people still had to use their
brains because memory, CPU cycles, and transmission bandwidth were at a
premium. Nowadays standards are forgotten and then reinvented (and
christened "standards" by anybody who cares to make them up), usually
without much thought behind them and often with disastrous consequences
(this is not aimed at the Unicode Consortium!).
Although this is rather far afield from current topics, I suppose it bears
restatement, since it is "old" and nobody remembers any more that there are
decades-old standards that tell us how to transmit any ISO-registered
character set in both the 8-bit and 7-bit environments, and to do so in a
simple and fairly efficient manner, using a choice (or mixture) of single
shifts and locking shifts. Yes, Kermit does this on 7-bit connections, and
it also adheres to the long-forgotten and now discredited dictum that only
standard "common intermediate representations" should appear "on the wire",
as opposed to any old character set one happens to be using. The
registration authority being the one and only internationally recognized
one, the ISO. Thus we may announce our character set using an ISO
registration number, allowing the recipient of text thus encoded to identify
it unambiguously and translate it to the local, perhaps proprietary and
unregistered, character set. This reduces an O(n^2) problem to an O(n)
problem and it promotes recognition of standards.
- Frank
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:37 EDT