% This looks like LaTeX; but it is really SGML.
% The "commands" are SGML elements, not TeX macros.
% A semi-colon at the end of a "command" forces command name termination.

\documenttype{article}
\baseloc{http://www.albany.edu/\tld;hammond/gellmu/mui/}
\date{October 22, 1999\\
     Last re-processing \today}
\author{William F. Hammond}
\title{The Idea of a Markup Interface for \abbr{XML}}
\surtitle{Univ at Albany: W. F. Hammond: The \abbr{GELLMU} Archive}
% Now let's squeeze the DVI into two U.S. pages
\latexcommand{\bsl;setlength\{\bsl;topmargin\}\{-30bp\}}
\latexcommand{\bsl;setlength\{\bsl;textheight\}\{696bp\}}
\nobanner % Suppress the "translated from GELLMU" advertisement
\nobaseprint % Since the document network location is spelled out
             %  explicitly below, don't provide it routinely for print.

\begin{document}

\section{The Value of a \latex;-Like Markup User Interface
         for \abbr{XML}}

A markup user interface (\abbr{MUI}) for an \abbr{XML} language
(formally \quophrase{\abbr{XML} application}) is a markup language
that admits a well-defined translation to the \abbr{XML} language.

Recent discussions in \urlanch{news:comp.text.tex} show that at least
two of us are thinking about \abbr{MUI}s for \latex;-like \abbr{XML}
languages.  Jonathan Fine has been working on a \emph{roff}-like
\abbr{MUI} that he calls \quophrase{Active TeX}, and I have been
working on a \latex;-like \abbr{MUI} that I call \abbr{GELLMU}.

As I use the acronym \abbr{MUI}, I do intend it to be reminiscent of
the acronym \abbr{GUI} for \quophrase{graphical user interface}.  Both
\abbr{GUI}s and \abbr{MUI}s have the intention of making life easier,
or perhaps at least more familiar, for some authors.  An \abbr{MUI},
unlike typical \abbr{GUI}s, has the possibility of giving the author
full and rigorous control over content.

Furthermore, an \abbr{MUI} in the style of a pre-existing
non-\abbr{XML} markup offers a convenient avenue for prototyping a new
\abbr{XML} language to model the markup practice in the pre-existing
markup.

Beyond that, it offers a route for conversion of legacy archives in
the pre-existing markup to \abbr{XML} languages with minimal human
intervention.

Please allow me to say a bit more about what I have in mind for
\abbr{GELLMU}.

\section{The Basic \abbr{GELLMU} Processing Design}

The things that I have on hand, aside from \latex; are:

\begin{enumerate}

\item  \abbr{GNU} Emacs, version 20.  (I believe that version 19 is OK.)

\item  James Clark's \quostr{nsgmls}, a part of his \quostr{SP}.

\end{enumerate}

My processing set-up is the following:

\begin{defnlist}

\term{Syntactic translation from \latex;-like markup to \abbr{SGML}}
\desc My Elisp processor, which can
be run interactively in \abbr{GNU} Emacs or in batch mode, performs
syntactic translation to convert \latex;-like markup to an \abbr{SGML}
language (formally, application).  The syntactic translator is largely
ignorant of command names.  Whatever command names are used become the
names of \abbr{SGML} elements.  There is a standard way to convert
multiple argument/option sequences.

This processing stage traps syntax errors.  (It will fail to detect an
even number of missing \quochar{\dol} characters; but this error will
show in the next stage.)

\term{Validating parse of the \abbr{SGML} language} \desc At the
validation stage the difference between \abbr{SGML} and \abbr{XML} is
significant if one wants to have \quophrase{math mode} be a global
toggle since that may be modeled robustly only using \abbr{SGML}
exclusions.

A validating parse is made using \quostr{nsgmls}.
Of course, the language definition, i.e., \abbr{SGML} application
definition, is crucial.  It is contained in an \abbr{SGML} declaration
and in an \abbr{SGML} document type definition (\abbr{DTD}).

The language definition that I am using is the heart of my personal
production system.  But I regard it only as didactic in the larger
scheme of things.  Others will certainly want things that I do not
want.

As far as it goes, it models \latex; rather closely in some ways and
rather loosely in other ways.  Where it departs from close modeling,
the reason is usually related to having a document structure that is
not print-centric.

The validating parse traps errors in language use.

\term{Down-translation to XML} 
\desc This is done with the program
called \quostr{sx} in the family of \quostr{SP} processors.

While it is possible to recover an equivalent \abbr{SGML} document
from the down-translated \abbr{XML}, it is not possible under the
\abbr{XML} umbrella to have as precise a language definition as
under the wider \abbr{SGML} umbrella.  That said, either form may
be run through a processor that serves to enforce a tighter language
definition.

\term{\abbr{SGML} processing}
\desc This can go anywhere that is sane for the language definition.
One can use any programming language, but it helps to have a basic
\abbr{SGML} library on hand.  I am using David Megginson's
\quostr{SGMLS.pm}, a Perl 5 library, and its interface
\quostr{sgmlspl}.

In my personal production system I routinely format \abbr{GELLMU}
\quophrase{articles} for both \latex; and \abbr{HTML}.  Invalid
\abbr{HTML} and error messages from \latex; represent bugs in my
processors, which I always repair as soon as possible.  There may be
box size complaints from \tex;; they represent authoring content
errors.  These two formatters work either on the \abbr{SGML} or the
\abbr{XML} version of an article.

At present my personal production language definition is not fully
up-to-speed for journal articles nor for translation to
XHTML-with-MathML, but it is serving me well for my classes.  It gives
me a sane way to have course handouts and web offerings in a
bullet-proof way from a single source.  At this time I simply cannot
assume that more than about a third of my students have easy access to
\abbr{PDF} readers.  So I feel constrained to give them simple
\abbr{HTML} with very limited pseudo-TeX for math.

Other formattings for \abbr{GELLMU} article that should be possible
include translations to (1) DocBook, (2) TEI, and (3) Texinfo, though
suboptimally so long as math is not available.  These are not small
jobs, and I may never undertake any of them.  (Any project that
undertakes to format an \abbr{XML} language in Texinfo should give
serious consideration first to modeling Texinfo as an \abbr{XML}.
It would also be desirable to minimize Info/\tex; bifurcation in
XTexinfo and to provide math for XTexinfo.)

If a time arrives when I can assume that three-fourths of my students
have out-of-the-box browsers for XHTML-with-MathML, then I should be
able to format the original \abbr{GELLMU} source directly for that ---
if by that time I am not able to squeeze it out of carefully-set
LaTeX-4.  (I exaggerate somewhat.  In principle, I need to provide
some math symbol declarations in the sources, and math symbol
declaration handling is not yet present in my set-up.)

\end{defnlist}

\section{About this document}

This document was prepared as a \abbr{GELLMU} \quophrase{article}.
A copy of the \abbr{HTML} version of this document is available
on the web at
\display{\urlanch{http://www.albany.edu/\tld;hammond/gellmu/mui/}}
along with a full list of anchored versions as follows:
\begin{menu}
\item \abbr{GELLMU} source: \urlanch{mui.glm}.
\item \abbr{SGML}, \quophrase{article} document type: \urlanch{mui.sgml}.
\item \abbr{XML}, \quophrase{article} document type: \urlanch{mui.xml}.
\item formatting in \latex;: \urlanch{mui.ltx}.
\item \abbr{DVI} made from \latex;: \urlanch{mui.dvi}.
\item \abbr{PDF} made from \latex;: \urlanch{mui.pdf}.
\item formatting in \abbr{HTML}: \urlanch{mui.html}.
\end{menu}

\end{document}