W3C home > Mailing lists > Public > public-dxwg-wg@w3.org > June 2019

Re: Do we really need a notion of profile "conformance"?

From: Thomas Baker <tom@tombaker.org>
Date: Wed, 19 Jun 2019 10:08:53 +0200
To: Rob Atkinson <rob@metalinkage.com.au>
Cc: Karen Coyle <kcoyle@kcoyle.net>, Antoine Isaac <aisaac@few.vu.nl>, Dataset Exchange Working Group <public-dxwg-wg@w3.org>
Message-ID: <20190619080853.GB15571@cicero.speedport.ip>
On Wed, Jun 19, 2019 at 06:06:21AM +1000, Rob Atkinson wrote:
> The argument seems to be something I dont understand at
> all, and have never seen an example of AFAICT: "most specifications are
> mutable" .

The Initial Example in [1], as shown in Figure 2, shows Profile X as a
profile of "Dublin Core Terms" (*), identified in the example as [2].

The specification at [2] has evolved, and continues to evolve, over a
long timeframe.  This is not unusual - think FOAF, Schema.org, even SKOS
and RDF.  For RDF namespaces, best practice turned out to be _not_ to
include versioning information in URIs. DCMI stopped doing it after
http://purl.org/dc/elements/1.1/ (the next one to be coined was
http://purl.org/dc/terms/). FOAF still uses http://xmlns.com/foaf/0.1/,
but changing the URI at this point would be quite disruptive with no 
obvious gains (other than aesthetic).


[1] https://w3c.github.io/dxwg/profilesont/#initial-example
[2] http://dublincore.org/documents/dcmi-terms/

(*) I have commented that the more correct name is "DCMI Metadata 
    Terms" but that is irrelevant to my point here.

Tom Baker <tom@tombaker.org>
Received on Wednesday, 19 June 2019 08:10:15 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:28:30 UTC