Dmitry,
> 12.07.2012 19:44, Paul Vinkenoog wrote:
> >
> > So... have we implemented the DT and CTE features correctly?
>
> I believe it has nothing to do with our implementation of DTs and CTEs.
> Create a view with gen_id and and select that field thrice - you will get the
> same result. And views are here for a few decades ;-)
>
> The "problem" is kinda conceptual. In regular selects and joins, only
> persistent fields compose an in-memory tuple, expressions are a separate
> thing and they are always evaluated per reference. If you reference them
> thrice, you will have three subsequent executions.
This seems to be wrong.
First, in the original example:
select u, u, u
from (select gen_uuid() u from rdb$database)
there is only one instance of the expression -- there are 3 references to the
*resolved* expression. Had the SQL been
select u1, u2, u3
from (select gen_uuid() u1, gen_uuid() u2, gen_uuid() u3 from
rdb$database)
Then I would agree that there are 3 references.
Sean
P.S. This subject would extend to how the engine currently considers all
functions as "variant" (the answer changes with each call), where most
functions/UDFs are "invariant". This causes optimization problems in cases
where the SQL reads:
SELECT ... FROM ... WHERE ColumnA = COS( :Param)
In this case, the result of the COS() function is a fixed value and thus the
optimizer should use any ColumnA index, but it doesn't. In reality, the SQL
should be treated the same as this:
SELECT ... FROM ... WHERE ColumnA =:Param
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
