Re: "Open Interchange"

At 20:16 3/6/97 +1000, Rick Jelliffe wrote:
>To give an example that might make more sense to British-derived readers:  
>
><!ENTITY Mac SYSTEM 
>	NDATA XML-char 
>		[ xml-role="CHARACTER"
>		xml-char-name="Ligature for Mac in Scottish Names"
> 		xml-equiv="Mac"
>		xml-class="letter" 
>		href="gttp://w3.org.uk/glyphs/scotland.font#3" ]
>
> which means: use the "M<sup>c</sup>" glyph if you can retrieve it,
> otherwise, for sorting etc. use "Mac".

Martin Bryan <mtbryan@sgml.u-net.com> wrote:
> Neat, but why reinent the world? XML will use DSSSL for presentation. Why
> can't it also use DSSSL for lexicographic ordering and glyph mapping?

Well, in terms of reinventing, this is closer to a TEI Writing Set
Definition -- see Harry Gaylord's paper on the issues surrounding these,
for example.  TEI WSDs deserve more attention from the wider SGML community
than they have received.

I certainly wouldn't want to make every DSSSL style sheet build in code
for specifying collation rules.

Sorting is different from presentation: St. Edmund sorts after Safeways and
before Sanctuary, but is presented with either a t. or an
    <overstrike>
	<superscript><smallcaps>t</smallcaps></superscript>
	.
    </overstrike>

or an underlined superscript T (an ordinal T, if you will), or whatever.

For text indexing, one might wnat to say that &mac; is to be indexed
as &mac; and also as its replacement content Mc. and also as "mac", or
onemight prefer to do query-time term expansion.  There is no need to
standardise that, and we don't have the implementation experience for
XML to make that appropriate anyway.

A "little language" like that for character entities is something that I
proposed at SGML OPEN last year, but the mailing list has been dead, not
least as I wasn't able to get to any SGML OPEN meetings since then.

Lee