Re: Deprecate http://www.w3.org/1999/02/22-rdf-syntax-ns# in favour of /ns/rdf# ??

On 2 December 2013 10:05, Christophe Guéret
<christophe.gueret@dans.knaw.nl>wrote:

> Hi everyone,
>
> No surprise at the level of opposition to deprecating the namespaces
>> (again, I point out in my defence that I raised it after someone asked
>> me about it; as a stickler for persistence I'm happy with that outcome).
>>
> As we recently talked about it, it could be that Phil refers to me there
> so feel free to blame me for the long mail thread ;-)
>
> I fully agree with many of the points that forked out of the main
> proposal, namely:
> * Make it easier to find out if something is in rdf: or rdfs: (and
> eventually why)
> * Provide different serialisations
> * Provide better multilingual support
> This would make everyone's life easier. Now.
>
> For the future I would still much advocate deprecating the old namespace
> and start using the new one. If I could give only one motivation, it would
> be that new vocabularies will be using the new namespace whereas the key
> ones will still be using an old (and somewhat confusing) location. From the
> outside this is not very consistent: several W3C vocabularies under
> different locations with no clear common design pattern, usage of dates in
> URI whereas most BP guides (rightly) say it is not a good idea, version of
> the vocabularies different than what the URI suggest, ...
>
> Furthermore, deprecating does not mean having to rewrite all the triples
> that are out there, and all the hard-coded namespaces used in every
> software. We can just set up redirects between the terms present in the old
> namespace to the one in the new one with a note suggesting not to use the
> old syntax any more. This may involve a bit more HTTP tips&tricks than just
> serving an OWL file but I don't see any big technological difficulty there.
> We can also keep this redirection until nobody uses the old namespaces any
> more...
>
> This will eventually happen as most of the RDF data out there is
> automatically generated from some legacy formats. As the maintainers of
> these datasets update their D2R/CSV2RDF/... scripts for using the new
> namespaces the usage of the deprecated ones will progressively fade away.
> Software also gets updated every now and then and new releases can safely
> come with new parsing capabilities. There is only data and software that do
> no see any update that will not switch to using the new namespaces... but
> do we really want to build a Web of Data on outdated data and abandonware ?
>
> Anyway, for the sake of clarity and end-user friendliness, I think we
> should keep this deprecation idea in the air while starting to work on
> content negotiation and multi-linguality
>

I fear this would be a recipe for mild disaster. Let's sketch a timeline.
Say W3C announces that rdfs and rdfs will "be considered equivalent to" the
URI http://w3c.example.org/rdfcore# in December 2013.

<FICTION>

Maybe January 2014 someone writes a patch for librdf's raptor, such that it
adds a mode to API and commandline of the Redland RDF parser suite such
that they can emit triples with the new URI instead of the original URIs,
or extra triples so we get both, "for maximum compatibility".

February 2014 someone forks https://github.com/semsol/arc2 with a similar
option, and also patches the SPARQL query engine to have option of silently
turning queries that use rdf: or rdfs: prefixes to be querying the new URIs
instead or as well. Dave Beckett says he'll wait to see if consensus
settles down before deciding what to do with the raptor patch.

March 2014, blog + twitter thread about whether it should also rewrite
INSERT statements https://github.com/semsol/arc2/wiki/SPARQL- spills over
into the Jena and Sesame mailing lists. No consensus there on exactly what
changes ought to be proposed. Andy circulates url to his post in this
thread.

April 2014, proposals on this list advocating for update to the recently
finalized RDFa 1.1 standard arguing that compliant RDFa parsers should
preferentially emit triples using the new namespace not the old confusing
ones, unless running in an explicit back-compatibility mode. W3C team
response is that RDFa and RDF WGs have already successfully completed their
chartered work, but that advocates for improvements should talk to their
Advisory Committee representative about proposing a new Working Group to
bring RDFa up to date here (or consider joining the Consortium if not
already a member). A rough draft charter is for this is debated on the HTML
WG lists, where it is suggested that the Microdata/RDF mapping document
should also be updated as an essential part of this work, but that the RDFa
Next Gen WG charter could also address the widespread desire for
convergence between Microdata and RDFa.

May 2014, heated debate during several ESWC 2014 linked data events on
whether the new namespace should have used # or /, and whether
http://w3c.example.org/rdfcore/ should redirect with a 303 or 302 to
/rdfcore#

June 2014, W3C advisory committee panel session on convergence plans. W3C
management state that it is important to address the needs of the RDF and
data linking communities but that coordination with key HTML standards is
critical, and that any new WG would need to carefully balance
backwards-compatibility considerations. An Adobe panel speaker notes that
several trillion bazillion JPEGs, PDFs etc contain embedded "classic"
RDF/XML and that it would be counter-productive for W3C to take any steps
that suggest these hard-to-update files are in any way invalid, e.g. by
encouraging old rdf: and rdfs: URIs to be treated as deprecated in query
languages and APIs. Another panelist suggests a Community Group to work on
a unit test suite for SPARQL implementations that would capture this
requirement. Lively discussion of relevance of SPARQL entailment regimes,
OWL and RIF is deferred due to 'lunch or beers' as the panel session is
overrunning its scheduled AC Meeting slot.

July 2014, Linked Europe Data 2030 EU project announced. Deliverables
include patches to Jena and Sesame to bring them up to date with new modern
RDF data linking standards such as  http://w3c.example.org/rdfcore#  Long
email threads sprawling across several mailing lists ensure.

August 2014, Email threads go quite for the northern hemispheric summer. A
draft "minimal" SPARQL 2.0 charter is circulated for discussion, addressing
"bugfixes and errata, RDF/S namespace versioning update and possible
convergence paths with XQuery. Also possibly a spec for extension functions
in a secure Javascript subset, if other work completes early. Chartered
deliverables include "RDF/S Namespace Versioning mechanism - Use Cases and
Requirements.". The WG charter MUST address the rdf:/rdfs: version handling
issue, and MAY address OWL and 3rd party namespace aliasing at the
discretion of the chair and staff contacts". The charter generates
suprisingly little discussion - perhaps due to the time of year.

September 2014, W3C announces "Beyond Maintenance Mode: rethinking RDF
usability for 2030 and beyond" Workshop to be held in December 2014, hosted
by Linked Europe Data 2030 project, workshop chairs include representatives
from Adobe, Jena, Sesame projects.

October 2014, RDFLib github repo merges a patch for partial
http://w3c.example.org/rdfcore#  support in its Turtle, RDF/XML and N3
parsers: debate in IRC, Twitter, github and email on whether bulk data
imports from the 3rd party Trig parser, or direct via API, should also be
canonicalized to use the new namespace, or whether this is even really
needed. Someone back-ports the initial RDFLib patch to work with earlier
versions, asks on mailing list how to make it available via easy_install.

November 2014, semantic-web@w3.org list thread in which we're reminded that
everything in /TR/ ought also to be updated so as to bring the usability of
W3C's RDF-related documentation into a more modern state. Discussion flows
as to whether a hypothetical Community Group to 'coordinate' this work
ought to also consider Oasis and IETF specifications. Somehow triggers a
crossposted thread on Mozilla lists "did we remove all the RDF/XML stuff
yet?".

December 2014, someone circulates some rules in N3 that show what a trivial
problem this would be if only tools supported basic semantic Web meshup
ideas. Heated discussion ensues.

... etc. </FICTION>

Just to emphasize, this is all so obviously fictional. There's no way we'd
get this much work done in 12 months.

If anyone thinks a more positive timeline is possible, do please sketch
one...

Dan