W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > May 2003

RE: typed literals and language tags - ugly parade

From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Date: Fri, 9 May 2003 11:40:53 +0200
To: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>, <w3c-rdfcore-wg@w3.org>
Message-ID: <BHEGLCKMOHGLGNOKPGHDEEJICBAA.jjc@hpl.hp.com>


This message summarises the disadvantages of each proposal (and the fifth option of doing nothing).

Doing nothing
=============

This leaves a language tag in the syntax of literals such as "2"@en^^xsd:int, which is (a) explicitly meaningless and (b) without rationale.
This is likely to lead to user and implementor confusion and possible interoperability problems.

> Option 1:
> 
> PROPOSE
>    XML Literals are as in the working drafts prior to November 2002, in 
> which it was not a typed literal, but a special sort of literal,
> with the changes made as a result of the reagle-01 and reagle-02 issues. 
> (i,e. exc-c14n performed in the syntax document)
>    Typed literals to exclude the language tag in the abstract syntax.
> 
> editors of Syntax, Concepts, Test and Semantics actioned to come 
> back with 
> text, based on current editors drafts, and last version before we 
> switched 
> to the rdf:XMLLiteral type, for the group approval.

This design was negatively received in earlier drafts.

With XMLLiteral as a distinct thing from typed literals then more implementors may choose not to implement it, causing interoperability problems between systems that support XML Literal, and ones that don't.


> > Option 2:
> > Literals can have both a type and a language tag if and only if
> > the type is
> > rdf:XMLLiteral, otherwise unchanged.
> 
> 
> PROPOSE
>   Concepts is changed to say that a literal can only have both a datatype
> and a language tag when the datatype is rdf:XMLLiteral.
>   Other editors to make consequential changes.
> 

This excerbates rdf:XMLLiteral being an anomolous datatype, in that the syntax is anomolous as well as the semantics.
This then has similar dangers  (to option 1) of a schism between implementors who can be bothered with it, and those who can't.


> 
> Option 3:
> PROPOSE
>    Typed literals, including XML Literal, to exclude the language tag in 
> the abstract syntax.
>    XML Literals to be refactored by deleting the <rdf-wrapper> text from 
> concepts and putting it into syntax (probably in para 7.2.17).
>    Add the following implementation note (or similar) to syntax.
>    Change NTriples in test cases to show explicit <rdf-wrapper> for all 
> XMLLiterals.

This is ugly in the syntax, and the <rdf-wrapper> hack becomes increasingly in-your-face. There is also a danger of non-backward interoperability with people who used to generate

<html>
<head></head>
<body>
<p>This comes from RDF</p>
</body>
</html>
Now getting
<html>
<head></head>
<body>
<p><rdf-wrapper>This comes from RDF</rdf-wrapper></p>
</body>
</html>

> > Option 4:
> > Language tag is simply dropped from all typed literals including
> > rdf:XMLLiteral
> >

This makes it awkward to embed xhtml inside RDF maintaining language information. Since this is an important use case we probably need to:
1) make sure that examples are included in syntax or the primer showing use of 
<span xml:lang="en"> to include langauge information inside a literal.
2) give clear warnings
3) alert I18N
Received on Friday, 9 May 2003 06:22:20 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:57:29 EDT