\centerline{\bf SGML Soldiers On}
\smallskip
\noindent
It is reassuring to know that the cold war still lives on
in these days of glasnost.  At Mark-Up '89 held in Gmunden,
Austria in April we were treated to young, glamorous  and
very earnest women reassuring us that \sgml{} would ensure
that the free world  remained free thanks to the
combat-readiness of the tanks, cruise missiles and
destroyers  and their accompanying truck loads of
documentation.

It is  apparent that \sgml{} afficionados can be divided
into two distinct  groups, the theorists and the
pragmatists. I have to say that it was this last group 
that made the conference worth attending. This
latter group, could  be further sub-divided into the
militarists and the commercial publishers and printers.  To
a great extent, the DoD suppliers have  to use \sgml{}
and its terrible progeny  CALS whether they like it or
not; their main aim is to find an easy way around  the
problem. The now famous Taft memo stating that CALS had to
be used for all DoD  contracts was widely discussed. (The
ultimate print-buyers revenge?). As a user  of a GenCode
based system (CAPS) I made it my task to search out the
{\sl voluntary\/}  fellow sufferers and I was particularly
interested to hear their experiences. As  usual, the Dutch
were there in force. I generally regard them as being
pretty well  ahead of the game in Europe and I must say it
came as a bit of a shock to me to  find that many of them
were still taking {\sc ascii} text and adding their own
mark-up  to it. In the UK there are some encouraging signs
with at least one other publisher/printer  (besides the OU)
developing an MS Word to \sgml{} program.

Elegant conversion from the authoring word processing program to \sgml{} is still very 
much the missing link. Avalanche Technologies have gone some of the way to solving 
it with FastTag which I believe is the XGML parser that is also used by DocuPro 
and ArborText. To my mind a `visual' system of hierarchy
recognition misses out  a lot of the subtleties of anything
other than simple text.

There certainly seems to be no shortage of `conventional' \sgml{} parsers available 
at a price, Sobemap's Mark-It and Yuri Rubinsky's SoftQuad being two notable examples. 
These both work as stand-alone word processors and would be great if you could persuade 
all your authors to use one.

Xyvision and Interleaf both gave demonstrations of their \sgml{} front-ends. Interleaf's 
was very neat but again it presupposed that the author was the person on the Interleaf 
keyboard. Doubtless it could take in ready-made \sgml{}. In terms of major installed 
gencode based publishing systems Compugraphic with CAPS was noticeable by their 
absence.

This conference was significant because for the first time people were able to report 
back on the results of actually using \sgml{} in competitive environments. The Oxford 
English Dictionary was offered up as an example of main-stream publishers using 
\sgml{} which makes a lot of sense when you consider updating and multi-media presentations. 
On the military side concern was expressed at how the number of standard DTDs was 
growing on a daily basis (they originally thought they would only need 3 or 4) and 
secondly at the amount of time it took to write a DTD. Noises were made about months 
rather than weeks.

The other worry voiced by the militarists was that of retro-fitting \sgml{}/CALS to 
existing documentation. How was it to be done and who was going to pay for it?

I have to confess that I found the discussions by the theoreticians on standards 
dry to the point of excruciating as well as a slight annoyance that they were apparently 
failing to recognize the amount of hard work being undertaken by users in order 
to breathe life into these standards. But I did come away from the meeting feeling 
that \sgml{} was actually being used by real people. When the conference comes to Europe 
again in two years time I believe that we will see that \sgml{} has progressed significantly 
across all EP applications.

\rightline{\sl John Feltham}