Re: Indirect Graph Identification

On Tuesday 17. January 2012 07.51.04 Sandro Hawke wrote:
> At this point, we've gone to Last Call on everything but this GSHP
> document, which I think we'll be done with very soon.  So, I don't think
> we'll be addressing anything not already addressed in the documents,
> unless we get public comments convincing us it's necessary to make our
> the material we have function properly.

At the end of the day, it is not the WG that needs convincing it needs 
changing, it is implementors who needs to be convinced they need to implement 
this protocol. I intend to implement most of it, but I have not yet seen any 
compelling arguments why I should implement indirect graph identification.
 
There was a thread starting at 
http://lists.w3.org/Archives/Public/public-rdf-dawg/2009AprJun/0306.html
and I was initially in favour of the 2b case, but has since not found any of 
the arguments forwarded there as compelling. 

Going back to the process, I'll note that there are numerous open comments on 
this in your comments tracker, 17 if I counted correctly, some of which may be 
the same comment though. One comment is from TimBL where he calls indirect 
graph identification a "distraction". While a response has been sent, no 
exhaustive treatment to these comments have been tried. 

Moreover, the this feature was something the WG did include in the "SPARQL New 
Features and Rationale" because it was thought to be easy (I was the editor at 
that time). It wasn't voted upon separately in at this early stage in the 
process either, as it wasn't amongst the features that had been submitted. 

This feature, especially the indirect graph idenfication hasn't been through 
the same process as the rest of the specs, and it is therefore my opinion that 
one should be conservative, do not add stuff that isn't clearly required. This 
to me means a simple, clear feature set with direct identification only.

Best,

Kjetil

Received on Tuesday, 17 January 2012 20:15:40 UTC