W3C home > Mailing lists > Public > spec-prod@w3.org > October to December 2000

Re: Thoughts and question about spec-DTD and stylesheets for translated specs

From: Eve L. Maler <eve.maler@East.Sun.COM>
Date: Mon, 20 Nov 2000 10:22:58 -0500
Message-Id: <>
To: Stefan Mintert <stefan@mintert.com>
Cc: spec-prod@w3.org
Hello Stefan,

At 11:35 AM 11/17/00 +0100, Stefan Mintert wrote:
>1) Separate annotations from specs and use XLink to link both together

For future reference, note that support is being built into Amaya that 
allows for the creation of third-party annotations using RDF and 
XPointer.  The only catch is that you don't have a choice of browsers when 
creating or viewing the annotations!

>I append my very first version of the trans-spec-DTD below. Any answer,
>comment or improvement is appreciated!
>At the moment I don't like that neither an instance of spec-DTD is an
>instance of trans-spec nor vice versa. I'd like to reach compatibility
>of some kind.

Your customization layer looks fine.  Compatibility will be a problem until 
translation markup facilities can be added to XMLspec, of course.  I'm 
collecting enhancement requests for V3.0, so your new elements may get added.

>The next step will be to change the XSLT-Stylesheet. Some of the string
>constants, such as 'Table of Contents', have to be changed. Is there a
>place in spec-DTD where to note the language in which the text is
>written? I don't think that 'langusage' is the right place, is it? What
>about an attribute to the spec-Element called 'xml:lang'. Of course I
>could add such an element to 'transheader'. But in my opinion,
><transheader xml:lang="de"> would mean that the contents of the
>transheader element is written in German. There's nothing said about the
>contents of the spec element.

Actually, the <language> element inside <langusage> is indeed meant to 
indicate the language in which the document is written.  I think this came 
originally from the TEI DTD.  These elements have been in the DTD longer 
than the xml:lang attribute has existed!

What do people think -- should I add xml:lang to the common attributes, or 
just to the top-level element, or what?  Note that the <scrap> element has 
a lang attribute that references the ID on a <language> element from the 
header.  (Hmm... While xml:lang is supposed to be able to identify "any 
natural or formal language", its values are constrained to come from RFC 
1766, which is meant for natural languages only.)  Should I keep this 
setup, or dump it?  If I dump it, what should I do on <scrap>?

My preferences: If I'm going for maximum cleanliness, I would want to 
remove the <langusage> structure entirely, put xml:lang in the common 
attributes, interpret xml:lang at the top level as the language for the 
entire document unless overridden lower down, and keep <scrap lang=> but 
turn it into a CDATA attribute that just contains a formal language 
name.  This is backwards incompatible, but could probably be handled in an 
XSLT conversion script in a reasonable manner.  (I doubt a lot of people 
even use <langusage> or <scrap lang=> today.)

Eve Maler                                          +1 781 442 3190
Sun Microsystems XML Technology Center    eve.maler @ east.sun.com
Received on Monday, 20 November 2000 14:34:12 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:55:07 UTC