On Thu, 11 Jul 1996 10:55:58 CDT Tilly Dutilh (INL, Postbus 9515, 2300
>Is there any consensus about the following matter:
>'rend' being a global attribute, highlighting attribute values such as
>'bold centered', 'italics', 'twice underlined' etc. are integratable in
>Should rendition attribute values preferably be integrated in tags?:
>1. <head rend=`bold centered'> Much about nothing </head>
>2. <hi=`bold centered'><head> Much about nothing</head></hi>
>3. <head><hi=`bold centered'>Much about nothing</hi></head>
For what it's worth, my personal view is that (1) is preferable,
at least in the case where the special rendition of the text is
(part of) what allows us to recognize it as an element of a
particular type (here: a heading).
Translating from a visual markup into an analytic markup is always
likely to be a trying experience, but I'm not sure why (1) would
be harder to produce from word-processor data than (2) or (3); in
each case, one has to recognize (a) that the element is bold and
centered and (b) that it is a heading.
(1) also has the advantage that in cases where all elements of a
given class are rendered similarly (e.g. all DIV1 HEAD elements
are bold and centered after a hard page break), the rendition can
be recorded once, in the header, and need not be repeated for each
occurrence of the element. This is somewhat harder if all changes
of font or format are tagged with the HI element.
-C. M. Sperberg-McQueen