W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > July to September 2009

RE: Personal Straw Poll about XMLLiteral

From: Seaborne, Andy <andy.seaborne@hp.com>
Date: Mon, 14 Sep 2009 15:43:19 +0000
To: Bijan Parsia <bparsia@cs.man.ac.uk>, W3C SPARQL Working Group <public-rdf-dawg@w3.org>
Message-ID: <B6CF1054FDC8B845BF93A6645D19BEA3693DEB2032@GVW1118EXC.americas.hpqcorp.net>

> -----Original Message-----
> From: public-rdf-dawg-request@w3.org [mailto:public-rdf-dawg-request@w3.org]
> On Behalf Of Bijan Parsia
> Sent: 14 September 2009 12:36
> To: W3C SPARQL Working Group
> Subject: Personal Straw Poll about XMLLiteral

Personal, and unrelated to this WG.

> We have a lot of triplestore folks in the group. While unscientific,
> it would at least give some idea of the feasibility of "fixing"
> XMLLiteral.
> So,
> 1) Does your RDF/XML parser canonicalize rdf:XMLLiterals?

Yes (it's required by spec)

> 2) Does your NTriples/Turtle/other alt-syntax parser canonical
> rdf:XMLLiterals?


> 3) Do you maintain the original syntax of your XMLLiterals, or only
> retain the canonicalized form?

If RDF/XML: lexical form after canonicalization.  Original not kept.
Other syntaxes: illegal XMLLiteral can be created (as for any other datatypes).

> 4) Would you object to changing the lexical space of rdf:XMLLiteral
> to include non-canonicalized wellformed XML?

Without a concrete proposal of what the change is to, it's hard to say.  Just relaxing the definition of lexical form is not enough.  What about CDATA? Entities? C14N has a list of things that get changed, some of which look quite sensible.

I'd like to understand how the changes interact with XML namespaces and language settings that are passed into the XMLLiteral during RDF/XML to produce a standalone lexical form.  If the RDF/XML parsing rule to canonicalize remains as-is, then I think that it gets messy but just removing it is also undesirable.  I do not see so much value in having more forms of RDF that can't be encoded in RDF/XML (not that it's impossible by other means already - another story).

> 5) Would you object to changing the RDF/XML parsing behavior to not
> canonicalize?

As above - there is a real issue with namespaces and languages that needs to be addressed so what's the alternative proposal in detail?  Would need to see that proposal to know whether to object or not.  And to discuss within whatever organisation I'm in at the time.

> The user version includes:
> 6) Would any of your applications break if the lexical space of
> XMLLiteral were widened?

FYI: Questions about XML literals (SPARQL or not) come up only very occasionally on jena-dev.

> 7) Would any of your applications break if the parsing of RDF/XML
> didn't canonicalize?

The test suite would break.

> Cheers,
> Bijan.


Received on Monday, 14 September 2009 15:44:18 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:00:57 UTC