Official Response to ISSUE-125 from RDF Web Apps WG

Hi Gavin, bcc: RDF WG,

Thank you for your public feedback on the RDFa 1.1 documents. This is an
official response from the RDF Web Apps WG to your issue before we enter
the 3rd Last Call for the RDFa 1.1 work this coming Tuesday. The Last
Call will last for 3 weeks, so there is still time for you to discuss
your concerns if we have not fully addressed them.

Your issue was tracked here:

ISSUE-125: Should CURIEs be more limited to not trigger on things like
http://example.com?
https://www.w3.org/2010/02/rdfa/track/issues/125

Explanation of Issue
--------------------

You wanted us to tighten up the definition of CURIEs. Specifically, you
requested this:

"Align RDFa with SPARQL and Turtle prefix names."

More to the point, you wanted to ensure that IRIs with schemes
containing "://" could not be mis-interpreted as a CURIE. That is, the
suggestion was to prevent "http://example.com" from being interpreted as
a CURIE if the "http" prefix was defined in the list of prefixes. You
also wanted us to bring the CURIE definition more in-line with the
Prefixed Names definition in TURTLE and SPARQL.

Working Group Decision
----------------------

The Working Group considered your suggestion and analyzed the various
consequences of such a change to the RDFa Core 1.1 specification:

http://www.w3.org/2010/02/rdfa/meetings/2012-01-26#Updating_the_CURIE_Syntax

We found that adopting the definition of Prefixed Names for CURIEs in
TURTLE and SPARQL would not adequately align CURIEs with Prefixed Names.
Far too many PN_LOCAL_ESC characters would have to be allowed in the
CURIE definition for the PN_LOCAL part, thus resulting in a definition
that is largely divergent from the definition of Prefixed Names in
TURTLE and SPARQL. That is, if we made the changes that you suggested,
we still wouldn't have good alignment with TURTLE and SPARQL.

The Working Group agreed that TURTLE, SPARQL, RDFa, and JSON-LD should
all have some sort of valid EBNF grammar for what constitutes a valid
shortened IRI. However, doing so should probably be the job of the RDF
Working Group. We would have probably adopted a universal definition for
shortened IRIs if one existed.

However, the Working Group found a real-world bug in the CURIE syntax
while discussing the issue. Facebook uses colons ":" in the reference
portion of their CURIEs, which has traditionally been illegal.
Additionally, the point you made about preventing certain types of '//'
IRIs from being detected as CURIEs was a concern for a few of the WG
members. In the end, the WG made the following changes:

RESOLVED: Make the change on the CURIE definition in RDFa Core 1.1,
according to Niklas' e-mail, allow for ':' and prevent the use of '//'
in the reference portion of a CURIE.

http://www.w3.org/2010/02/rdfa/meetings/2012-01-26#resolution_2

Niklas' e-mail can be found here:

http://lists.w3.org/Archives/Public/public-rdfa-wg/2012Jan/0067.html

The production rule we adopted is Option C.

Feedback
--------

Since this is an official Working Group response to your issue, we would
appreciate it if you responded to this e-mail and let us know if the
decision made by the group is acceptable to you as soon as possible.

-- manu

-- 
Manu Sporny (skype: msporny, twitter: manusporny)
Founder/CEO - Digital Bazaar, Inc.
blog: PaySwarm vs. OpenTransact Shootout
http://manu.sporny.org/2011/web-payments-comparison/

Received on Saturday, 28 January 2012 19:39:57 UTC