Re: invitation to tomorrow's WG meeting

Hi Peter,

Please see responses below

Irene
> On Mar 7, 2017, at 6:18 PM, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:
> 
> Here is my currently incomplete analysis of the current major problems with
> SHACL.  Please redistribute to the working group.
> 
> 
> 
> There remain several major problems with the definition of SHACL:
> 
> 1/ SHACL-SPARQL depends on pre-binding.  There has never been a suitable
> definition of pre-binding for SHACL.  The current definition has a couple of
> technical problems that need to be fixed.  The current definition
> depends on subsituting fresh variables into SPARQL queries but there is no
> demonstration that this is well behaved.  The current definition also
> changes the meaning of some SPARQL queries.  Pre-binding is also needed for
> quite a bit of the informative text in the SHACL Core parts of the document.

This is a topic of a separate e-mail you have sent earlier today, correct?
> 
> 2/ (FO) There is no requirement that SHACL implementations have a mode that
> rejects syntactically-invalid shapes graphs or shapes graphs that contain
> constructs that the implementation does not handle.  Without this
> requirement users do not have a guaranteed way to determine whether their
> shapes graphs will work the same in other SHACL implementations.  It is
> possible to create conforming SHACL implementations where users have no idea
> whether any of their shapes graph will be processed the same way on other
> SHACL implementations.  Syntax checking is quite easy for SHACL Core at
> least so there is no good reason not to require it.

Wiki page: https://www.w3.org/2014/data-shapes/wiki/FO-3:No_requirement_to_reject_graphs_with_invalid_shapes <https://www.w3.org/2014/data-shapes/wiki/FO-3:No_requirement_to_reject_graphs_with_invalid_shapes>
> 
> 3/ The syntax of SHACL does not make sense.  (FO) Syntactic constructs that
> are useful or unobjectionable have been removed.  The syntactic rules are
> written too broadly, making useful shapes graphs syntactically illegal.
> (This problem is made much worse by the lack of required syntax checking.)

Wiki page: https://www.w3.org/2014/data-shapes/wiki/FO-1:Removing_features_from_node_shapes <https://www.w3.org/2014/data-shapes/wiki/FO-1:Removing_features_from_node_shapes>
> 
> 4/ (FO) The new sibling disjoint stuff looks to be well-defined but very
> unusual.  I haven't finished my analysis of all the weird corner cases.
> There will need to be lots of tests to cover these cases.

Wiki page: https://www.w3.org/2014/data-shapes/wiki/FO-2:Disjoint_siblings <https://www.w3.org/2014/data-shapes/wiki/FO-2:Disjoint_siblings>
> 
> 5/ I haven't had a chance to fully check out the new definition of
> validation reports but it appears that it is much too loose.

I am afraid this is currently not actionable. Please let us know how much time you will need to finish you analysis and provide comments.

> 
> There is also the problem that several comments have not had a substantive
> response from the working group and a few haven't had any response at all.

As Sandro said in his e-mail:

If you have made earlier comments that you don't consider as having been addressed, please tell us again.

> 
> 
> peter