Re: “SHACL Minus SPARQL”

On 3/4/2015 15:25, Jose Emilio Labra Gayo wrote:
> And if one party of the discussion is very interested to include those 
> implementation details in the spec, I would ask them to be in a 
> separate section at least.

This has already happened - the SPARQL bits are separated into

     http://w3c.github.io/data-shapes/shacl/#sparql

The generic template declaration vocabulary is currently independent of 
SPARQL.

Some people in the SPARQL camp seem to be concerned that we would risk 
balkanizing the SHACL community if we allow alternative languages. 
However, you are proposing a controlled sub-set of SPARQL expressions, 
or maybe XPath, so would it work for you if we define SHACL such that:

1) Every well-formed constraint or template must have a sh:sparql query
2) Every constraint or template may have additional executable bodies in 
other languages

This makes sure that *all* models are executable against the reference 
SPARQL engine, while still supporting engines that do not have a full 
SPARQL processor. You can then enrich the default core templates with 
something like shx:xpath and your engine can transparently flag errors 
if it encounters files that only have sh:sparql and no shx:xpath.

Thanks,
Holger

Received on Wednesday, 4 March 2015 05:42:35 UTC