Path: utzoo!attcan!uunet!world!decwrl!bacchus.pa.dec.com!deccrl!shlump.nac.dec.com!galvia.enet.dec.com!killian
From: killian@galvia.enet.dec.com
Newsgroups: comp.text.sgml
Subject: Re: Umlaut vs. Diaresis (?)
Message-ID: <1990Oct9.163803@galvia.enet.dec.com>
Date: 9 Oct 90 15:41:42 GMT
Sender: newsdaemon@shlump.nac.dec.com
Reply-To: killian@galvia.enet.dec.com ()
Organization: Digital Equipment Corporation
Lines: 60

To: manfred@swi.psy.uva.nl ()
Cc: 
Subject: Re: Umlaut vs. Diaresis (?)



> Does SGML provide a way of specifying an 'Umlaut', as is used in German, in
> contrast wit a 'diaresis', used in many languages, such as the Dutch
language.

In general, yes!  There are two possible ways of doing this but the choice 
of which depends on whether you want to enter the accented character in text 
(ie: content) or markup.

In the more common case where you want to enter the accented character in 
text, SGML would advise the use of an SDATA entity (system specific data 
entity).  For example, &ouml; could be used to enter a small 'o' with an 
umlat, and &odiar; could be used to enter a small 'o' with a diaresis.

Of course these entities need to be declared in the scope of the Document 
Type Definition. This is normally done for a complete collection of such 
entities and transportability is enhansed when user communities can agree 
on, and standardise, such entity sets.  'ouml' is present in the ISO Latin 
1 entity set published as an informative annex to the SGML standard; its 
description includes 'o' diaresis, so this particular entity set will not 
solve your problem.

In addition, an SGML system must be able to correctly interpret the 'o' 
diaresis entity reference when the parser finds it in the text. For example, 
your SGML typesetting system must be able to translate the declared value of 
the 'odiar' entity to the correct glyph shape.

The other solution, which would also allow the use of the accented character 
in markup (eg: tag names), is to use a character set that has a code position 
for the accented character (a different code position than the similar 'o' 
umlat).

SGML is character set independant, in that the SGML declaration (before the 
Document Type Definition, but absent from most SGML documents) allows the 
identification or definition of the document character set. 

Of course, the SGML parser must be able to accept the SGML declaration (not 
every one does) and the SGML system must be able to accept (eg: typeset) text 
in that character set. Defining your own character set is not always a smart 
thing to do.

I have also seen some unconventional solutions to your problem. One such 
solution involved defining a special element (tag) that was used to enter 
accented characters. For example: <special>odiar</special>. 

Again, this element would have to be defined in the scope of the Document 
Type Definition and the SGML system would have to be capable of translating 
the 'odiar' text into the required accented character.

My recommendation is to use the SDATA entity solution if the accented 
character is not required in markup.


Regards,
	Aidan