Not to pick on Dan's particular message.
It was just the last one to provide me with a handy response point.
Looking back, the original 1990-1991 argument was between "Just send 8 bits"
with no tagging of the content, vs "tagging and bagging" to deal with
the fact that 8-bit at that time would break too many systems, and
the problem of needing to tag the types of text characters that were
contained in RFC822 message bodies.
But, as there were two problems to solve (8 bit SMTP ) and (tagging
RFC822 content), the decision was made (after a very long and very
heated argument) to split that problem according to the boundaries of
SMTP and RFC822.
In fact, the two problems were and still are totally distinct.
The RFC822 solution group produced MIME with Quoted-Printable as its
charter specified, to assure interworkability as soon as possible.
The 8-bit folks did not do anything then or since to require 8-bit carriage.
So, it is time for a working group to solve this problem.
It is not going to help anyone to argue about the past, as in fact,
it was very rational at the time, and one side of the problem was
ignored as soon as quoted printable relieved the immediate pressure
for an SMTP solution.
So, now is a good time for some constructive discussion of ways to
solve the charset problems, hopefully with some useful charset
tagging tools for the non-RFC-822 parts of the SMTP envelope. (But I
do not mean to specify either the work or the results at this point.)
It is indeed unfortunate the the DRUMS WG did not find a way to solve
the 8-bit SMTP problem, but now is the time to get to it. There
should not be too much 7-bit SMTP code to fix by now.
At 8:35 AM +0000 4/1/02, D. J. Bernstein wrote:
Claus Faerber writes:
> D. J. Bernstein <djb(_at_)cr(_dot_)yp(_dot_)to> schrieb/wrote:
> > I'm not saying that Quoted-Printable had no short-term benefits for its
> > short-term costs. I'm saying that, viewed from our long-term perspective
> > eleven years later, the failure to require 8-bit transparency was an
> > amazingly stupid decision.
> From our present perspective, that's true. Back then, it might
> have been the best solution.
No. The failure to require 8-bit transparency was inexcusable. Every
approach that failed to require 8-bit transparency could have been
dramatically improved. Consider, for example, these three approaches:
(2) Quoted-Printable plus ``you must handle unencoded 8-bit data too'';
(3) pure 8-bit without this 7-bit garbage.
Whether or not you understand that #3 would have been better than #2,
surely you understand that #2 would have been vastly better than #1.
> Further, remember that the first MIME standards date back to 1992.
> Back then, Unicode was brand-new and UTF-8 only came with the 2.0
> version in 1996. Without UTF-8, you just could not even think
> about using Unicode in message headers; and without Unicode, you
> could not solve the charset-labelling problem.
Get your facts straight. First, UTF-8 was introduced years before 1996,
although under another name. Second, even without knowing about UTF-8,
people _were_ thinking about Unicode in headers, and proposed several
workable approaches. Third, even without Unicode, there were several
solutions to the character-set labelling problem.
Anyway, none of this is relevant to IETF's acceptance of 7-bit garbage
in 1991, and none of it is relevant to IETF's acceptance of 7-bit
> The movement towards UTF-8 everywhere is quite new.
Again, get your facts straight. The ``movement'' started with X-Open,
Rob Pike, and Ken Thompson a decade ago. RFC 2277, requiring UTF-8
support for all text on the Internet, is four years old.
---D. J. Bernstein, Associate Professor, Department of Mathematics,
Statistics, and Computer Science, University of Illinois at Chicago