Re: multiple value shape?

Folks, you both have important points and we all need to be careful what 
we state here, to not give wrong impressions to the outside world.

First we need a resolution on whether other languages than SPARQL can 
serve as extension languages. I have opened

     https://www.w3.org/2014/data-shapes/track/issues/60

for that purpose. Once we decide to do that, then I would like to take 
back my statement that "SPARQL is always a fallback" and suggest the 
more general "Extension languages such as SPARQL or JavaScript are 
always fallbacks". This should cover the use cases that Arnaud talks about.

I want to note however that we are on a slippery slope here, and we need 
to be clear about the role of the Core Vocabulary. Stating that users 
cannot use SPARQL features because not every platform supports them is 
extremely dangerous, because it would lead to a situation similar to the 
"Don't leave OWL DL" scaremongering. It would basically discourage 
people from using the very power of SHACL only because *some* 
implementations don't want to support SPARQL, or cannot support SPARQL. 
But these implementations may turn out to be minorities. We cannot 
anticipate this yet, but discouraging SPARQL use will serve as a 
self-fulfilling prophesy.

The other dimension of the discussion is that the more features we add 
to the Core language, the more complex it becomes for users and 
implementers. In the end, it might become just as complex as SPARQL, and 
we have exactly the same situation over again: many users wouldn't be 
able to use it (too hard to learn), and not every implementation would 
support all SHACL Core features. My answer to all this is that a 
flexible extension mechanism will go a long way.

On the political dimension, I also understand Arnaud's desire for the 
SPARQL camp to listen to the ShEx camp, and make sure that their 
favorite use cases such as Qualified Cardinality Restrictions can be 
expressed via the core language (and thus the Compact Syntax). I am very 
open to this, assuming we can agree on a realistic definition of QCRs. 
But at some stage we need to stop adding features to the Core and let 
the extension mechanism and the template libraries do their job. It is 
not our job to re-invent a poor-man's variant of SPARQL or JavaScript 
and expect all vendors to implement it, and all users to learn it. 
People already familar with SPARQL will prefer SPARQL, and people 
already familiar with JavaScript will prefer JavaScript.

Thanks,
Holger


On 5/30/2015 1:32, Arnaud Le Hors wrote:
> "Peter F. Patel-Schneider" <pfpschneider@gmail.com> wrote on 
> 05/29/2015 07:57:34 AM:
>
> > ...
> > > Should be possible but not necessarily practical. -- Arnaud  Le 
> Hors -
> > > Senior Technical Staff Member, Open Web Technologies - IBM Software
> > > Group
> >
> > Possible, and thus available as a fallback.
>
> Sorry but I find this line of reasoning silly so I'm not going to 
> answer any further.
> --
> Arnaud  Le Hors - Senior Technical Staff Member, Open Web Technologies 
> - IBM Software Group
>
>