images, notations, backends

[Adam Di Carlo <adam@xxxxxxxxxxx>]

I've been thinking about a rather difficult problem -- image handling
in SGML.  As you now, different backends like different image formats.
For instance, if you're producing DVI or PS with JadeTeX, you use EPS
format.  If you're producing PDF (with PDFJadeTeX), your figures need
to be in PDF format.

Reading Goldfarb's "SGML Handbook", there seems to be a rather
amorphous and foggy notion that you can use "Dual-role" public text
(system and public declarations), such that the system identifier
points to a program which can intepret the output (looking at 6.4.4 on
page 188):

  The external identifier of a data content notation, for example,
  could include both a public identifier of a publication that
  describes the notation, and a system identifier of a program that
  will be called to interpret data that conforms to the notation.

My problem is that I would like a build system which, based on the
output, pre-processes figures. I can teach the system what figures to
create based on what output format is desired.  But then I also have to tell the SGML source what image file suffix to expect.

I was wondering if there's a more standards based way to do it, or if
it makes sense to have OpenJade (or anotehr SGML processor) look for
external data with NOTATION attributes and run a certain program with
certain arguments or some such (sdc used to do this).

-- 
.....Adam Di Carlo....adam@xxxxxxxxxxxxxxxx<URL:http://www.onShore.com/>


 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist