[
Lists Home |
Date Index |
Thread Index
]
Greetings,
On Tuesday, October 8, 2002, at 11:48 AM, Rick Jelliffe wrote:
> It does not mention the editors they use, which would be useful,
> but most use TeX or the ISO 12083 DTD which allows
> shortreferences.
This seems like a good point to chime in. The American Physical
Society publishes high quality physics journals and has
been using electronic typesetting technologies since troff and eqn
first made an appearance. In the late 80's we switched to TeX for
creating camera copy (and TeX is still the preferred tool of
authors). Now we use two different vendors. One (listed on
Rosenblum's survey) uses a Xyvision system in which markup-only
macros are added by the compositors. The Xyvision files are then
post-processed into SGML which can then be modestly repurposed.
What is really missing is an "SGML-first" process which would ensure
that the SGML file that we have for archiving. The ISO 12083 math
is part of the SGML file, but it can't be used to re-typeset the file
reliably and it can be difficult to work with.
Our other vendor is an SGML-first shop, but they had been using
ArborText
for the composition and typesetting. This uses the AAP math and a TeX
backend for the actual typesetting. The problem here is that it is
often necessary
to drop down to PI's to feed extra info to the TeX backend. For
instance, to
create a long radical sign extending over a complex formula, some like
<?Pub Eqn="\sqrt{"?> is put in and similarly the closing brace is added
via <?Pub Eqn="}"?>. This means that in order to reuse this material,
a lot of reverse-engineering of PI's is necessary, again defeating
the goal of having archival SGML that can be extensively repurposed.
About 5 years ago we had a chance to work with SoftQuad to try
and do "SGML on the Web" (to tie into another recent xml-dev thread).
However, this project failed and we have been patiently waiting for
XML technologies to progress to the point where we can create
archival XML first, typeset it directly in a high quality manner, and
then make the XML available directly on the web with minimal
transformations
and styling. This has always been "one year aware".
However, I think we are finally getting there. Our vendor that had been
using ArborText now uses it just as an editor - the ArborText math
is then converted to MathML and it is formatted quite capably by 3B2.
The XML files we get back are finally what I would call "archival" in
nature with little extraneous information, largely self-documenting,
and have some hope of actually displaying in a mainstream web browser.
The main issues now are fonts for the special glyphs STM publishers
use and getting MathML display up to snuff. The first is being taken
care of with the STIX fonts project - www.stixfonts.org. This should
be done by the end of 2003. We have also been able to add to Unicode
many of the glyphs. The issue here is that many of these have been
put on plane 1 and most software doesn't grok that yet. It is also
proving to be quite complicated in coming up with a packaging scheme
for the fonts themselves so that they will have the widest degree of
usage across applications and operating systems.
As for MathML display, I think we are almost there. David Carlisle
is much more able to comment on that. Anyway, I think it has
been a long road, but that there has been a steady progression
toward our pipe dream of making richly marked up content available
directly to end user researchers without being constrained by
print mode of dissemination.
Rosenblum (and Lapeyre) have also been working with publishers,
the National Library of Medicine, and Harvard University to come
up with a very nice archival DTD for STM publishing which I think
will have a very good chance of being widely adopted. It is notable
that although the APS's new XML DTD and this new effort were
largely independent, the final designs show a lot of convergence
which I find encouraging.
To sum up, I think SGML has given us some degree of comfort, but
the way it is practiced falls far short of what we really envisioned
when
we moved to SGML. XML is just getting to the point where I think
that we really are just one year away...
Cheers,
Mark
Mark Doyle
Manager, Product Development
The American Physical Society
doyle@aps.org
|
