🚀 go-pugleaf

In article <4yyNVB4w165w@blues.kk.sub.org>, kosta@blues.kk.sub.org (Kosta Kostis) writes:
> henry@zoo.toronto.edu (Henry Spencer) writes:
> 
> > However, vjs's main point is correct:  there is no way to mandate updates
> > to every piece of relevant software everywhere, which means that backward
> > compatibility is (unfortunately) fairly important.  
> 
> There is no way to mandate updates especially if there is no "need"
> for that. As long as we don't use 8-bit, nobody will ever change
> "faulty" software.
>
> >                                                    If the purpose of
> > sending the message is to communicate, rather than to score political
> > points, you want to be pretty sure that it's intelligible to the audience.
> 
> It's not political, it's just sensible.   :-)
> 
> > It will not be possible to forget about the bit-handicapped until their
> > numbers are vanishingly small.  That won't happen soon.
> 
> I don't agree. It depends on us. Since the intro of VT 2xx terminals
> in 1983 nine years went by and many people are already equiped with
> 8-bit terminals and PCs, Macs, NeXTs, Workstations of various brands
> now. All these are capable of displaying 8-bit characters.
> 
> Those who don't need 8-bit, like people who will never read/write
> anything but english text, won't be bothered, but it would really
> help the rest of us.
> ...


All of that depends on the assumption that you control both ends and
all intermediate forwarders that might ever touch your mail.  That
assumption can be valid in a private network and sometimes among
communities in the Internet.  Those Internet communities must be either
very tightly knit or distinctly less than modern themselves, to avoid
having any MX records that point outside.

If, because of the vicissitudes of MX records, UUCP map entries, and so
on, your mesage happens to go to a machine running old software, then
your message is likely to either be stripped to 7 bits or lost entirely.

There are many 100,000 machines in the Internet whose users think they
do not need 8-bit text.  The users and administrators of those machines
will actively resist changing their software only because your mail
suffers when it passes through their machines.  If your mail contains
control characters in headers that cause sendmail on their machines to
crash, they will blame you and the vendor, and many will still actively
resist changing their systems.  Even if you bought the machines, paid
the salaries of the system administrators, and paid for the maintenance
of those machines, you would find they would still resist changing, as
the U.S.  Department of Defense found during the "host table" conversion.

Consider another historical example.  Most people agree that "rabid
rerouting" of email is impolite, unnecessary and just plain bad.  Those
who disagree have only some weak efficiency arguments.  Rabid rerouting
has always caused mail to be misdirected, bounced, and lost.  Many
people have complained publically and privately.  Still, as far as I
know, the big rabid rerouters are still at it, 5 or so years after the
practice was universally condemned.  The rabid rerouters include sites
that at least once upon a time forwarded a lot of mail.

What Henry wrote was literally true.  "There is no way to mandate
updates."  Not in many places in the world.  Certainly not in cultures
like that of the U.S. where many people make a point of defying the
"mandates" of those without genuine power.

Waiting for all or even only most machines to be upgraded is hopeless.
I personally wish it were otherwise, if only to stop hearing about
problems in products that I worked on years ago.  Most people just do
not think upgrading a systems that works well enough for them is a good
use of their time.  People who to try to force them to upgrade their
systems are either ignored or castigated.  That may not be sensible or
convenient, but it is a fact.

How much European and Japanese mail still passes through the U.S. with
its large number of recidivists?


Vernon Schryver,  vjs@sgi.com