Re: I added some comments to the wiki page for ISSUE-37

On 11 Dec 2012, at 20:27, "Wilde, Erik" <Erik.Wilde@emc.com> wrote:

> hello all.
> 
> On 2012-12-11 5:21 , "David Wood" <david@3roundstones.com> wrote:
>> Hierarchies of containers are necessary in my opinion, but I can't see a
>> good reason not to allow (MAY) a graph structure that would map cleanly
>> to the RDF model. 
> 
> there's no need for a MAY if we just define a protocol for a flat model.
> atompub does not talk about nested collections, because it is outside of
> the scope of interactions you need for basic functionality. many
> implementations have created conventions/additions that allow collections,
> but from the protocol level standpoint, it doesn't matter at all.

AtomPub did not speak about many things because they were working with very weak
tools. 

1. They were working at the syntax level and had limited themselves to
using atom syntax  (Atom Entry & Feed) which itself had a lot of arbitrary 
restrictions.  People had spent years discussing syntax issues - e.g.: where 
do you put the src of some content: should it be an element, or an attribute, 
etc... so that in the end any  small and minor thing they wanted to do would 
take them ages. Any talk of re-opening a syntax issue would of course send 
shivers down the editors spine.

  => for LDP there are no such syntax issues. We just need to define a URL for a 
concept, or relation and we are done, across any number of syntaxes we wish.

2. They had no modelling skills at all. Even simple UML diagrams did not seem
to be readable by most members present. It's IETF don't forget, and in this
case everyone (who could blog it seems) could participate, however little he 
knew or understands about anything. In any case they certainly did not have 
the semantic web logic at their disposal  to be able to think logically about 
consequences of what they were doing. As a  result you'd get an inferno of 
heated incomprehension.

  => we don't have that, we have some of the best logicians who worked on
  the semantics. If we want to see what the consequences of a statement, 
  we are not in the dark, as the Atom people were. We have standards we
  can refer to that can help move things along very far

3. We can look at the atom extensions and incorporate them, since we now can see
 what they have available. It's just a matter of semanticising some of these things.

4. They had no query language either.

I can go on and on, on how this simple syntax that was just a minor variation of
dublin core, ended up taking years to get to an XML version. A huge noise for nearly
no interesting result in my view. A little cottage industry then grew up trying to
fix things with new RFC specs here and there, which I never felt it worth spending
the time to follow. 

But the way you are going about things is very familiar to the Atom group: make
sure you get people to remove useful tools from the discussion, so that instead
of doing something in a day, it then takes years.

 

> 
>> Does anyone have a good argument against containers within containers (I
>> hope not).
> 
> i think the best argument is to say that for a basic platform, there's no
> need to address this question, unless we want to build interactions around
> that notion.
> 
> cheers,
> 
> dret.
> 
> 

Social Web Architect
http://bblfish.net/