W3C home > Mailing lists > Public > public-rdfa-wg@w3.org > January 2011

Writeup of yesterday's discussion on RDFa default profiles

From: Ivan Herman <ivan@w3.org>
Date: Fri, 28 Jan 2011 14:48:56 +0100
To: W3C RDFa WG <public-rdfa-wg@w3.org>
Message-Id: <CE0A4FB1-73AA-4C99-B8A9-7402D9643B3D@w3.org>
(This is relevant to ISSUE-78 and ISSUE-73 )


Manu asked me to give a short write-up of the discussion we had yesterday on the default profiles. Although we made minutes, it is probably a good idea to extract things, including maybe aspects where the scribe (yours truly:-) could not properly catch...

1. Default profile = an RDFa profile at a well specified URI (see later) and whose declaration in an RDFa source is not required. A default profile's priority is the lowest of all profiles referred to in the RDFa file. That means: if element 'e' is the top level XML element of the XML content, or the <html> element of the HTML5 content, then

- if the @profile attribute does not appear on element 'e', then, conceptually, the @profile="URI" is added before processing the DOM tree
- otherwise, if the @profile attribute's value is List-of-URI, then the attribute is modified, again conceptually, to be @profile="URI List-of-URI" before processing the DOM tree

2. We have one default profile file, valid for all RDFa host languages. Host languages cannot define default profiles.

3. The default profile may contain term and prefix definitions. There is currently a discussion in the group as for how the content of the profile file would evolve. In particular, options on the table are:

3.a: A default profile is defined by the RDFa WG and is published alongside the RDFa 1.1 specification. The URI of that profile is fixed and its content is then frozen. New (future) versions of RDFa may define their own profile but those should reside at a different URI-s.

3.b: A default profile is defined by the RDFa WG and is published alongside the RDFa 1.1 specification. The URI of that profile is fixed. The Working Group will also develop a mechanism whereby the default profile can be regularly updated by extending it with new prefixes and terms. There should be a clear and documented policy how that should happen, and that policy should work beyond the existence of the RDFa Working Group. 

I believe there is an agreement on item #1 (we never explicitly discussed this, but I believe what is there is the common idea we have on what a default profile is). There is also an agreement on item #2 (we also have a Resolution at [1]).

For 3.a vs. 3.b

- Pro 3.a: stability means that implementers can 'hardwire' the default profile in their implementation
- Pro 3.b: the evolution of the the profile file is bound to new versions of RDFa, which in turn is bound (per W3C process) to establishing a Working Group. This mechanism is way too heavy to answer the quick evolution of vocabularies, and that would make the very existence of profiles irrelevant

In the case of an 3.b approach, the frequency of profile updates was also discussed. Whatever the mechanism of registration is, a way of reducing the load on implementations, as well as network is to set the frequency of official new releases low; during the discussions we heard 6 months, 1 year, or 2 years.

As for the URI-s for the default profile, that of course depends on whether we go 3.a or 3.b. In the case of 3.a, the URI may reflect the RDFa version it is bound to, ie, something like


If we go towards 3.b, this may not be o.k. A structure reminiscent to the http://www.w3.org/TR might work, though, whereby each profile has a dated, and the stable, dateless URI refers to the latest release. That would allow RDFa authors to stick to one particular profile in case they need that (by putting it into their source explicitly).

Finally, both in the mechanism and in the choice of the URI we may want to take into account that the RDF working group may consider re-using the same (or similar) mechanisms for, eg, a future version of turtle. (B.t.w., that may mean that the vocabularies we use for the prefix and term definitions should not be in a namespace that bears 'rdfa')

I hope I could catch everything...


[1] http://www.w3.org/2010/02/rdfa/meetings/2011-01-27#resolution_1

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf

Received on Friday, 28 January 2011 13:48:54 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:05:23 UTC