Re: ldp-ISSUE-15 (sharing binary resources and metadata): sharing binary resources and metadata [Linked Data Platform core]

On 4 Oct 2012, at 22:13, "Wilde, Erik" <Erik.Wilde@emc.com> wrote:

> hello andy.
> 
> On 2012-10-04 12:28 , "Andy Seaborne" <andy.seaborne@epimorphics.com>
> wrote:
>>> I worked on the Atom spec. And at the time I put together an AtomOwl
>>> ontology http://bblfish.net/work/atom-owl/2006-06-06/AtomOwl.html
>>> I am not sure it is that good.
>>> But it could be a basis to see how close ldp ontology is to Atom.
>> I'm all for adopting any other useful technology.  That said, I also
>> think container-in-containers and standardised container creation are
>> import because so much is about information organisation.

yes, I very much would like container in containers!

> 
> atom is out there, ready for us to use. in terms of container management,
> that has been left out of AtomPub but could be added (there are several
> options for that, but they all have been implemented in various places,
> and the advantages and drawbacks are pretty well-known, i'd say. for
> hierarchical containers, there are proposals (and people have implemented
> that as well), but again there's no standard. but both issues could be
> tackled by extensions to AtomPub, allowing us to start from a standardized
> starting point and add what we need to add, based on our use cases.

The job of the Atom working group was never finished, because they dropped
the semantic part. I think it is good to look at what they have done, 
semanticise their work. Then see how it maps to ours. It should be easy then 
to create a GRDDL for atom that maps it to the work here and everybody can 
be happy. 

But I think the group will make a lot faster progress than the atom group
ever could because we are thinking at the semantic level, and so we don't 
have to get lost in issues that were entirely related to xml syntax and
to the weakness of tools of people implementing them.

Where there are differences they could be interesting. 
  One thing I noticed is that the Atom spec is in fact a way to publish data and
metadata about a resource simultaneously. It is not clear that this is such a good
idea - especially for resources with binary content - xml encoding binary content
is just not very nice. The content in atom can be a pointer, so that the binary 
resource can then have its separate url. I would rather start from this.

So yes, we should be looking at atom, but I think one can do a lot better
and a lot more simply once one allows oneself to think semantically.



Henry


> 
> cheers,
> 
> dret.
> 
> 

Social Web Architect
http://bblfish.net/