Re: Addressing ISSUE-64 and ISSUE-65


Drawback of the (DB-generated) constraint-name based approach is the 
nondeterminism (depends on platform, history) which makes it difficult 
to specify SPARQL queries in a platform-independent way.
Benefit is that length of the IRI is not an issue.
(In practice length may not matter too much with the proposed naming 
involving parent and child table/col-sequences. But, the constraint-name 
based approach could always be a fall back.)

- Souri.

Richard Cyganiak wrote:
> On 10 Aug 2011, at 19:34, David McNeil wrote:
>>> Souri, all: what are the cons of having the constraint name in the IRI?
>> It can be a generated identifier which can be painful to look at,
> Human-generated identifiers can be painful to look at too. I wouldn't consider this a problem.
>> and not-deterministic (i.e. it is not a function of the DDL).
> Good point. Yes, that *is* a problem to some extent. It makes writing test cases for the DM  impossible unless you explicitly specify the constraint names in the DDL.
> Best,
> Richard

Received on Wednesday, 10 August 2011 19:00:28 UTC