Re: why ISSUE-45 won't really work ( POSTing to an LDPR appends content to the resource )

On 9 Jul 2013, at 02:17, Steve Speicher <sspeiche@gmail.com> wrote:

> On Tue, Jul 2, 2013 at 2:40 PM, Henry Story <henry.story@bblfish.net> wrote:
> I think I found a reason why POSTing to an LDPR as an APPEND is not as useful as one
> might have thought. (I know the issue is closed but until now I thought it was one
> of those issues that could be re-opened)
> 
> An advantage of POSTing a graph to an LDPR as an append operation would have been
> that it could take care of SPARQL UPDATE's INSERT DATA method easily, and could have
> removed the need for it. There are good reasons one could have for only allowing
> addition of triples, never removing triples.
> 
> 
> LDPR doesn't prevent this, POST is undefined on LDPR.

I know, but this is what ISSUE-45 was discussing.

>  
> But then one has the problem of how one would append triples to an LDPC. One still
> seems to need PATCH for that.
> 
> Yes, PATCH seems like the right choice.
>  
> Or are there very good reasons for why one should really never append triples to an
> LDPC?
> Never append triples?  We don't prevent that.

Well if POST were a way of appending triples to an LDPR then the current spec could be seen
as  preventing the appending of triples that appear in the post since we in fact append a triple to a 
new resource created by the POST.

> Honestly I'm not sure what you are pointing out is against what the spec recommends (or remains silent).  It sounds like everything you want to do is allowed.

I am trying to see if there is a way to see a POST of triples on an LDPC as being the same thing as a POST on an LDPR.
ISSUE-45 suggested that POST on an LDPR should append triples. An LDPC is a subset of an LDPR, so if
POSTing meant append triples, then it should mean the same when posted to an LDPC.

IF we were using Trig - which is not so far out - then with the notion of quotation that comes with it, one could argue
that POSTing to an LDPC appends the quoted triples of the newly named graph that is formed in the post. 

<> ldp:created <m1> .
<m1> { <#me> a foaf:Person;
                         foaf:name "Jack" . }

So one could bring the two together by thinking of a POST as an appending of triples
to the resource created, which sometimes in the case of non LDPCs, is just the same
resource as the one POSTed to.

All I am trying to do here is see if there is justification for a POST as append.

Now the advantage of POST as append, is that one does not introduce *deletion* semantics
that a PATCH inevitably includes. Deletion on the web is to be avoided as much as possible,
I think because ADDing new montonic relations does not make other documents
false (as long as they don't use closed world reasoning). So keeping systems consistent
is a lot easier if one only allows additions.


Henry

> 
>  - Steve Speicher
> 
> 
> Henry
> 
> Social Web Architect
> http://bblfish.net/
> 
> 
> 

Social Web Architect
http://bblfish.net/