[Mulgara-dev] another select bug (MGR-36)

[Andrae Muys]

On 21/11/2006, at 1:58 PM, Life is hard, and then you die wrote:

>
> I just filed a bug report, http://mulgara.org/jira/browse/MGR-36 , for
> another problem with a simple query that we discovered recently.
> While we think we have a workaround, it would be really helpful if
> somebody more knowledgeable could take a quick look and give us a hint
> as to what is triggering the problem so we can check all our other
> queries.

Looking at the query, both should return the result of the first  
query.  The only significant difference between them that I can see  
is that the second query includes a join involving a non-union  
compatible disjunction.  I suspect this is another case of the same  
problems you were reporting a couple of weeks back with NUC- 
disjunction.  Clearly something I'm going to have to have a close  
look at once I finish the current work on transactions.

Yes the tuples algebra is a semi-ring, so naturally the distributive  
law should hold, and equally clearly in this case it is not - you're  
right in identifying this as a bug.

Andrae

P.S.  Note that while both queries include NUC-disj, only the second  
then attempts a join with it.

-- 
Andrae Muys
andrae at netymon.com
Principal Mulgara Consultant
Netymon Pty Ltd