RE: [FrameSGML] Re: [FM+SGML] Forcing an attribute to have a value

["Lynne A. Price" <lprice@xxxxxxxxxxxx>]

At 02:47 PM 4/6/00 -0500, Hassen Vince D Contr OC-ALC/LAKRT wrote:
>Lynne A. Price wrote:
>	>the r/w rules were never intended to be a general
>	>purpose transformation language
>
>I'm curious.  What were the R/W rules intended to be?

Vince,
  I sometimes describe SGML as an "input-only standard". The
SGML standard specifies what an SGML document is, but not how
it is to be processed. In the general case, it takes a full-blown
programming language to render an SGML document. For example,
I once heard about a FrameMaker+SGML application that supported
an SGML document in which one element with declared content
of EMPTY was to be rendered as a table populated with data
scattered throughout the document. An FDK client was used to
gather the data and construct the table.
  There is no reason for the r/w rules to provide this much
capability. After all, the FDK already provides a general-purpose
programming language interface into FM+SGML during import and
export. Nevertheless, there is a middle ground between arbitrary
processing and a direct everything in the SGML document must
correspond exactly to something of the same structure in
the FM+SGML representation. Furthermore, there's nothing in
an SGML document to indicate which elements correspond to
special FM formatting objects such as graphics, tables, markers,
variables, and cross-references. The r/w rules address both
issues. They provide a simple way to make some common, easily
specified transformations (e.g., changing element or attribute 
names in the SGML and FM+SGML view of a document) and they
allow the developer to indicate the type of FM object represented
by an SGML element. The FDK provides the ability to handle
more complicated transformations.
  Obviously, opinions can vary on what a simple transformation
is, and even on whether a particular type of data rearrangement
occurs frequently enough to be worthwhile in the r/w rules.
I do not at all suggest that the r/w rule capability is perfect
and couldn't benefit from enhancement. However, resources would
be better spent in making other enhancements to FM+SGML than
in extending the r/w rules to a general purpose programming
language.
	--Lynne

Lynne A. Price
Text Structure Consulting, Inc.
lprice@txstruct.com
http://www.txstruct.com

** To unsubscribe, send a message to majordomo@omsys.com **
** with "unsubscribe framers" (no quotes) in the body.   **