Re: shapes-ISSUE-11 (S9 impossible): S9 is about existing but unspecified values

The spirit of the original story is just the inheritance of further
restrictions from "not required" to "required".  For a Contract in general,
the end date is no required, but for a Bond, it is.   This is even
summarized in the original wiki (I think by pfps) in the sentence, "The
larger requirement here is restriction refinement in subclasses"


>From a Shapes point of view, I think it is correct to say that it is
impossible to say, "An end date exists, but we haven't specified it" (that
sounds like an open-world statement, and I think Shapes makes close-world
statements).  If that is the aspect that is deemed "impossible" here, then
I concur.


My point in telling the story may be subsumed elsewhere; it is just that it
should be possible to extend a shape that has no requirement for something
being filled in (in this example, that's end date) by adding a requirement
(e.g., in a sub-shape) that it does have to be specified (and perhaps has
other constraints, e.g., end-date should come after start-date).

This story was intended as an example; I would not be surprised if the
principle for which it is an example has been stated more generally
elsewhere.


Dean





On Wed, Jan 14, 2015 at 4:26 PM, Peter F. Patel-Schneider <
pfpschneider@gmail.com> wrote:

> The problem is that some of S9 appears to be asking for a constraint that
> states that a contract can have a property value for its end date.  This is
> different from the ICV constraint provided for bonds, which requires a
> specified value for the end date.
>
> What would a constraint that required that it was possible to have a value
> for a property look like?  What would it mean?
>
> peter
>
> On 01/09/2015 03:41 PM, Dean Allemang wrote:
>
>> The issue that S9 is about (as Peter outlines here
>> https://www.w3.org/2014/data-shapes/wiki/User_Stories#S9:_
>> Contract_time_intervals)
>> is that it should be possible to add constraints in subclasses where none
>> existed above.  The ICV example on that page seems to address this quite
>> directly; at one level, it doesn't represent a constraint, while at the
>> next
>> level down, it does.  The meaning of this might not be fully clear - I
>> have
>> added a paragraph to the description on the Stories page that I hope
>> clarifies
>> it.  Basically, it seems to me that the ICV code has got it right.
>>
>> So, far from being impossible, it seems that there is a solution presented
>> right on the wiki.
>>
>> On Thu, Dec 18, 2014 at 7:10 PM, RDF Data Shapes Working Group Issue
>> Tracker
>> <sysbot+tracker@w3.org <mailto:sysbot+tracker@w3.org>> wrote:
>>
>>     shapes-ISSUE-11 (S9 impossible): S9 is about existing but unspecified
>> values
>>
>>     http://www.w3.org/2014/data-shapes/track/issues/11
>>
>>     Raised by: Peter Patel-Schneider
>>     On product:
>>
>>     Story S9
>>     https://www.w3.org/2014/data-shapes/wiki/User_Stories#S9:_
>> Contract_time_intervals
>>     appears to require constraints that state that a property has a value
>> but
>>     this value is not specified in the graph.  Do any proposals cover this
>>     requirement?  Is this a constraint at all?
>>
>>
>>
>>
>>
>>