- From: Michael Kifer <kifer@cs.sunysb.edu>
- Date: Thu, 23 Apr 2009 12:49:50 -0400
- To: Dave Reynolds <der@hplb.hpl.hp.com>
- Cc: Sandro Hawke <sandro@w3.org>, public-rif-wg@w3.org

This is quite different then. You want to supply a binary predicate as a parameter to the builtin. I don't know how really frequent something like this is (for comparing lists), but I feel it is not. Furthermore, because of the restriction that predicates and function symbols are disjoint, this would be illegal syntax in BLD and Core. michael On Thu, 23 Apr 2009 09:09:01 +0100 Dave Reynolds <der@hplb.hpl.hp.com> wrote: > Michael Kifer wrote: > > >> 2. We got rid of deep-equal because you can just check for list > >> identity. But it seems to me there will often be lists of data > >> values, and data values generally need to be compared with the > >> appropriate equality predicate for their type (numeric-equal, > >> datetime-equal, duration-equal, XMLLiteral-equal, > >> compare+numeric-equal 0, text-compare+numeric-equal 0, etc). I > >> suggest bringing back deep-equal for when you want comparison on > >> data values to be done using datatype equality instead of identity. > >> (This does suggest having a literal-equal or data-value-equal > >> predicate; literal-equal(x,y) is the same as deep-equal(list(x), > >> list(y))). > > > > Data type constants that are equal will compare as equal. That is, > > List(1.0 2.0) = List(1.00 2.00). > > > > We don't have identity, by the way. Identity is not logically definable. > > We have previously used "identity" in the XSD sense of "same value in > the value space", which is exactly how our equality is defined. > > I believe what Sandro is pointing out is that for several of our > datatypes "same value in the value space" is less useful than the XSD/ > XPath notion of equality. For example: > > "2000-03-04T23:00:00+03:00"^^xsd:dateTime > and > "2000-03-04T20:00:00Z"^^xsd:dateTime > > are not identical under XSD 1.1, not equal in terms of our equality > semantics (they represent different values in the value space), but they > are equal in terms of real world time instants[*] and in terms of the > xsd equality algorithm which is exposed through our builtin > pred:dateTime-equal. > > So when comparing ground lists then the datatype-specific equality > predicates rather than the RIF "same value in value space" equality > semantics would probably be desirable. > > Dave > > [*] At least under the common sense, but technically false, notion of > absolute time. > > > > > -- -- michael

Received on Thursday, 23 April 2009 16:54:21 UTC