Ned,
Yes, I received a response prior to the publication of RFC 1341. However,
since that time, I have continued to receive messages only to the effect
that richtext will be subject to review "later". As far as I can remember,
the first time that was supposed to take place was in November 1992, but
nothing happened. I regret to point out that I have reason to remain
skeptical: The "past" isn't over until this pattern changes.
The reason that richtext has not been widely discussed by people germane to
document representation languages is that it was mostly perceived to be an
insignificant part of a mail format standard, and that the working group
was perceived to be vastly under-interested in this topic. I'm pleased to
see that you seem to be willing to discuss this seriously in a different
forum, and I shall wait impatiently for this to actually happen.
On the issue of SGML in particular, I'm regret to note that you're a little
confused, and exaggerate the problems because of this. To be specific, a
DTD is not a bunch of declarations in SGML useful only to me and to SGML
parsers; it's a "Document Type Definition" (as defined in ISO 8879 SGML),
only the formal part of which is specified in SGML. The rest of the
document type definition includes a specifiation of the semantics of the
elements (which richtext lacks, and which I tried to point out in my draft
comparison document). Moreover, a number of "application conventions" can
be specified in these (informal) specifications, and a few follow naturally
from working within the Internet mail environment.
I brought up the availability of parser materials to make it reasonably
simple to implement SGML-conformant software because one of the major
arguments that you presented against "full" SGML conformance was,
precisely, lack of software that understood the format (despite the fact
that there existed no software that understood richtext at the time this
statement was made, and that the ARC SGML parser materials were released in
June 1991, a whole year prior to publication of RFC 1341, and also prior to
the above statement).
Previously, there weren't any visible overlap between mail and document
representation languages because mail used to be only headers, and the body
used to be no more than "lines of text", subject to ad hoc user agreements.
This shows in MIME, as most problems have been solved by adding headers to
formalize the user agreements. richtext was the only major departure from
this "header paradigm", and it appears that only a few people were
interested in this "shift", let alone noticed it. It is particularly
evident in the proposed solutions to the character set issue, for instance.
This demands a change in focus, away from most of the "mail" issues, and
into the realm of "markup languages" (of which richtext is a kind). This
again demands that we attract different people, and don't alienate them by
making false and exaggerated claims about complexity to quell discussion
and the bringing forth of ideas and proposals.
It seems that SGML is a very difficult concept to describe to people, and
I've tried to condense a description of it into something like this:
For all sorts of information objects intended for human perception
(documents) that can be regarded as consisting of kinds of information
containers defined by their types of content (including data (text),
and nested information containers in a hierarchical structure), SGML
provides both a language to describe each kind of information container
(element), in terms of its content and the allowed sub-elements and
their order and occurrence patterns as well as attributes that each
instance of an element can specify to provide element-level meta-
information; an abstract syntax used to linearize the element structure
into one or more named character string objects (known as "entities");
and information about character sets, concrete syntax, and other
meta-information needed to interpret a particular SGML document.
SGML stands for the "Standard Generalized Markup Language", and the heavy
generalization of markup languages involved makes it possible to use it for
a wide variety of purposes that do not immediate yield themselves to be
described as "markup languages". "Information structuring for interchange"
has been used to describe one of SGML's uses. I think this is what mail
and MIME are all about. (However, for mail purposes, we already have a
"header paradigm" and departing from it only when we have good reason to,
makes it less useful to specify all of MIME in SGML, although that could
easily be done.)
To conclude: I think a separate working group and mailing list should be
set up to discuss content types for Internet mail, and a structured, richer
text format in particular.
Best regards,
</Erik>
--
Erik Naggum ISO 8879 SGML +47 295 0313
Oslo, Norway ISO 10744 HyTime Watch this ^ space
<erik(_at_)naggum(_dot_)no> ISO 9899 C Memento,
terrigena
<SGML(_at_)ifi(_dot_)uio(_dot_)no> ISO 10646 UCS Memento,
vita brevis