Re: fixing http://w3c.github.io/data-shapes/shacl/

On 09/07/2015 11:16 PM, Karen Coyle wrote:
> 
> 
> On 9/7/15 9:15 AM, Peter F. Patel-Schneider wrote:
>> I feel that the version of http://w3c.github.io/data-shapes/shacl/
>> that I reviewed (dated 10 August 2015) has three fatal problems.
>> - It does not reflect working group consensus.
>> - It provides a poor and misleading description of SHACL.
>> - It has  multiple severe technical problems.
>> Fixing of any of these problems requires significant changes to the
>> document.
>>
>> Fixing the worst aspects of these problems requires at least the following
>> changes:
>> - Writing an introduction that reasonably describes SHACL.
> +1
>> - Replacing the example that uses controversial aspects of SHACL.
> Which example is this?
Example 1.

>> - Using a well-defined vocabulary to describe the major SHACL notions.
> +1 - presumably we could list the terms and arrive at a consensus
>> - Correctly describing the relationship between SHACL and RDFS and SPARQL.
> It would be nice to hear what you think this is.
I just sent out a message on the former.

There is no indication that SHACL cannot be implemented in SPARQL.   (This
point may be moot because of the changes being proposed for the document.)
The use of SPARQL for the meaning of SHACL is very poorly discussed.

>> - Discussing recursive shapes.
>> - Fixing or removing the UML diagrams.
> Specificially, what is wrong with them?
I don't think that they correctly state multiplicity, but I am not a UML expert.

>> - Fixing the description of how violations are reported.
> ... and what is wrong with it?
The document indicates, for example, that violations inside negations are to
be reported.
> 
> Personally, I have no idea what you mean by these criticisms, and need quite a
> bit more explanation. I really liked your proposed wordings, so I'm hoping
> there is something to be done in these areas. But without more detail this
> does not help me understand.

Most of the details are in my long review.

> kc
>>
>>
>>
>> Peter F. Patel-Schneider
>> Nuance Communications
>>
>>
>

Received on Tuesday, 8 September 2015 11:08:29 UTC