Re: LDP Agenda for December 17, 2012, with a list of issues to be closed

In my responses to Ashok's questions I am assuming that by 'container' we are referring to strong compositions of resources rather than weak aggregations of resources, as defined by issue-25.

On 14 Dec 2012, at 23:51, Ashok Malhotra <ashok.malhotra@oracle.com> wrote:

> To make progress, I think we need to agree on a few fundamental features of the model.
> Let me list what I see in an effort to help the discussion on Monday.
> 
> Can a LDPR belong to more than one container?  If yes, then we need to specify how a LDPR that already exists is added to a container.
In my opinion, absolutely not (sorry David). It defies all previous thinking on composition and would make resource management an absolute nightmare. On the other hand there should be no restriction on the kinds of aggregations (collections) we can construct purely with RDF. However, I don't really believe there is disagreement on this issue, but that we're really talking about two different things - compositions v aggregations - the very point of issue-25. If Ashok is asking about aggregations then of course a resource can be a member of multiple aggregations/collections - if it can be expressed in RDF then it is valid.
> 
> When a container is deleted do are all its members deleted also?  I think we have agreement that all members may be deleted or they may not be deleted.  We need a mechanism to specify which: perhaps an attribute on the container called DeleteMembers yes/no
If composed members are not deleted along with the container then it begs the question as to what containers are for in the first place. It is central to the concept of composition that when the container is deleted its contents are also deleted. Again, if we're talking about aggregations/collections expressed in RDF then my answer is reversed; deleting one node in an RDF graph should have no consequences on any other node.
> Can a client create containers?  I believe the WG sentiment is running to YES
Yes - but perhaps only as a child of another container so that there is always a well-defined root container.
> Can containers be members of other containers?  I believe the WG sentiment is running to YES
Yes - but I could swing both ways on this. My mental model for this is the file-system model with a hierarchical URI naming convention, though I could easily go for a flat named-graph model that doesn't naturally nest.
> What did I miss?
> 
There is, lurking in the detail, another implicit container model which is the RDF representation/model associated with a given LDPR. This is an RDF graph that may contain a number of RDF nodes, one of which is hopefully the resource itself. When we delete or overwrite this representation then any resources within it will naturally disappear. This is a flat, graph-centric view of containers that could easily be extended to meet most of our requirements of containers - except perhaps for nesting.

Steve.

> All the best, Ashok
> 
> On 12/14/2012 2:12 PM, Arnaud Le Hors wrote:
>> 
>> Hi Askok, 
>> I'd be more than happy to discuss the model but I'm really not sure we can have a very constructive conversation without a concrete proposal to look at. 
>> 
>> For now, Erik created a wiki page [1] and invited people to try and describe what model they think we should have. I'm sorry to see that very little has been added to it (I know you did). Without this, again, I'm afraid we can spend a lot of time talking without getting anywhere. 
>> 
>> I also think this is a very important issue (as you may remember I was one of the first people to raise it in the first place) but looking at our list of open issues it's clear to me that some are quite independent of the model and we could at least make some progress in addressing those. 
>> 
>> Let's discuss this on Monday's call and see if we can come up with a plan on how to move forward. 
>> Regards. 
>> 
>> [1] http://www.w3.org/2012/ldp/wiki/ISSUE-37 
>> --
>> Arnaud  Le Hors - Software Standards Architect - IBM Software Group
>> 
>> 
>> Ashok Malhotra <ashok.malhotra@oracle.com> wrote on 12/14/2012 01:41:49 PM:
>> 
>> > From: Ashok Malhotra <ashok.malhotra@oracle.com> 
>> > To: public-ldp-wg@w3.org, 
>> > Date: 12/14/2012 01:43 PM 
>> > Subject: Re: LDP Agenda for December 17, 2012, with a list of issues
>> > to be  closed 
>> > 
>> > Arnaud:
>> > On Monday's call can we have a discussion of the model?
>> > I think we need to settle this basic issue before we go further.
>> > If you like, I can send you a list of questions or you can find them in the
>> > mail thread. 
>> > All the best, Ashok 
>> > 
>> > On 12/14/2012 1:22 PM, Arnaud Le Hors wrote: 
>> > Hi folks, 
>> > The agenda for Monday is available. Per last week's discussion, it 
>> > lists 5 specific issues to be discussed, with a proposal on how to 
>> > close them. Check them out! 
>> > http://www.w3.org/2012/ldp/wiki/Meetings:Telecon2012.12.17
>> > --
>> > Arnaud  Le Hors - Software Standards Architect - IBM Software Group