Re: more sorting test cases

Jeen Broekstra wrote:
> 
> FYI I have checked in a few new sorting test cases, and have amended 
> test case 3 slightly (to be more reflective of the current spec).
> 
> - test case 3 sorts lexically on URIs where possibly an unbound value
>    occurs (amended from original: only 1 unbound now occurs).
> 
> New cases:
> 
> - test case 4 sorts on datatyped (integer) literals.
> 
> - test case 5 sorts on two variables: first on untyped literal name
>    (lexically), then on datatyped literal. The query here is identical
>    to the third example for ORDER BY in the current WD (see section
>    10.1).
> 
> - test case 6 sorts on a mixed result of literals and URIs.
> 
> - test case 7 sorts on typed literals with different types (float and
>    int).

I have executed this test cases successfully and checked the results properly 
now.  The hardest part was processing the indexed result sets without using a 
query to access the indexes in order :-)

> 
> AFAICT this about covers most specified cases. I've come up with a 
> number of other test cases as well (mixed bindings of untyped and 
> typed literals, bnodes, etc.), but for most of these the spec does not 
> specify a specific ordering.

Ordering between mixed items is determined but ordering within them may not be.

"""
(Lowest) no value assigned to the variable in this solution.
Blank nodes
URIs
RDF literals
"""

A total ordering isn't appropriate as Pat pointed out - it depends on what 
datatypes the processor understands.

	Andy

> 
> Jeen

Received on Friday, 29 April 2005 15:00:55 UTC