Re: New Specification Published!

The arguments are getting a little blurry to me here, and I find
myself on the side of all of them :-)

1. Especially if we make it all the way to a W3 Recommendation, the
social issues arising from the cachet of W3 seem to be large and favor
being in a W3 namespace.

2. The problems arising from changing namespaces are exactly(?) two:
   a. Scenarios in which there are data using terms in both namespaces
will have unintended different semantics and integrability. For
example sparql queries asking something about <ns1>hasBody X will not
find data with <ns2>hasbody...
  b. Resolution and dereferencing may require server-side support  for
redirection, or else complete abandonment of the older namespace.

3. Once we decide to change the namespace at all, there is no extra
pain for choices we make about its details, but it should remain
stable thereafter.

4. (Strong opinion) Developers chasing a moving target in the stages
from which OA is now emerging are fully aware of the costs of
namespace changes, and also have the programming competence to write
the database and other scripts necessary to recraft existing data.
(Or, if they don't this will be a good learning experience for them
and they should thank us.  Hah, hah, just serious. ).

This all leaves me with this:  +0.75 for changing it, but with great
care about its stability and +0.25 for leaving it as is.

Bob Morris

On Wed, Feb 6, 2013 at 6:49 PM, Robert Sanderson <azaroth42@gmail.com> wrote:
> Hi Raphaël,
>
> On Wed, Feb 6, 2013 at 11:19 AM, Raphaël Troncy
> <raphael.troncy@eurecom.fr> wrote:
>
>>> 1.  We've had that namespace published for quite a while now, and
>>> people are implementing using it.  While they may have to change the
>>> code to update to the new specification, if they were only using the
>>> basics, they might not have to.
>>
>> Being one who implemented it, I can re-assure you that this will not be a
>> problem. I'm curious though how many oa annotations one can find out there,
>> published on the web? Any study on this?
>
> No study and not all of these provide globally accessible systems yet,
> but here's a list of pretty well known institutions that are or have
> implemented just off the top of my head:
>
> British Library
> Bibliotheque Nationale de France
> Oxford University
> Stanford University
> Meertens Institute
> John Hopkins University
> St Louis University
> Drew University
> New York University
> Brown University
> University of Queensland
> Yale University
> Harvard University
> University of Illinois at Urbana-Champaign
> LANL
> ...
>
> Yes, it's not a big deal to go in and change the namespace
> definition... but I'd like a better reason for it before we try to get
> all these developers to do it :)
>
>> I also agree that if you plan to make a change (and I understand that you don't want), then it is better to make it earlier than later.
>
> Yes, definitely!
>
>> My personal feeling is that the community is
>> just now actively looking at the new spec and that there is still a time
>> window to change it, but I might be very wrong and under estimate how deep
>> OA has already been implemented by the DL or other community.
>
> It's a bit deeper than that, but maybe not so deep as we'd end up with
> parallel universes.
>
>
>> So, there are two aspects: i) the fact you indeed spell it completely
>> 'openannotation' instead of 'oa' and ii) also that you use a path in your
>> ns.
>
>
>> Hence, should we expect that http://www.w3.org/ns/openannotation/ is
>> itself a small ontology that contains modules defined in
>> http://www.w3.org/ns/openannotation/core/ and
>> http://www.w3.org/ns/openannotation/extensions/, etc.
>
> Yes, I could buy the argument to lose /core/ now that we don't have
> anything in /extensions/ .  On the other hand, it's probably good for
> the future to have the possibility of /extensions/ if we need it.
>
> Is openannotation/ an ontology? I don't think that it needs to be,
> just like /ns/ doesn't need to resolve to a super-ontology either.
>
>> No hash in your ns either? People might wonder why you do differently than other vocabs.
>
> >From the set of namespaces that we use, not including our own the
> tally looks like:  / has 5, and # has ... 5 :)
> Unless there's a W3C best practice that we should follow that we don't
> know about?
>
> Many thanks!
>
> Rob
>



-- 
Robert A. Morris

Emeritus Professor  of Computer Science
UMASS-Boston
100 Morrissey Blvd
Boston, MA 02125-3390

IT Staff
Filtered Push Project
Harvard University Herbaria
Harvard University

email: morris.bob@gmail.com
web: http://efg.cs.umb.edu/
web: http://wiki.filteredpush.org
http://www.cs.umb.edu/~ram
===
The content of this communication is made entirely on my
own behalf and in no way should be deemed to express
official positions of The University of Massachusetts at Boston or
Harvard University.

Received on Thursday, 7 February 2013 08:52:42 UTC