Re: Oracle 9i features?

From: Serge Rielau <srielau_at_ca.eye-be-em.com>
Date: Sun, 08 Feb 2004 18:59:35 -0500
Message-ID: <c06ihg$65t$1@hanover.torolab.ibm.com>

Hans Forbrich wrote:

> Daniel Morgan wrote:
> 

>>Serge Rielau wrote:
>>
>>
>>>I shouldn'thave given the SQL Server example. Should have known only
>>>that part would be addressed, my bad.
>>>I don't think Pascal's question was whether Oracle can create a view.
>>>Let's presume a more complex example to illustrate.
>>>Be somecol a shape and morecol another shape. now use intersect(comecol,
>>>morecol)->shape.
>>>overlap() is an expensive operation. The idea is to keep the computed
>>>result persistent for faster access to shift the computation from the
>>>frequent select to the rarer update at the cost of storage.
>>
>> >
>>
>>>Cheers
>>>Serge
>>
>>Well don't let Joe Celko see this or he'll go ballistic. There is almost
>>no excuse in a relational database for storing computation results.
>>Perhaps in a datamart storing metadata but not in an OLTP system. I'd
>>solve the problem in Oracle 9i or 10g by writing a function, native
>>compiled in C, the compute the result on-the-fly.
>>

> 
> 
> Ya-but ... the competitor does it, so it must be OK <g>
> 
> However, the shape example above is kind-a irrelevant since that is
> aleady available in intermedia locator and spatial.  So we can say -
> been there, done that (using the solution you outline).
> 
> /Hans

Did I state for "illustrative purposes" when I referred to shapes? Yes I did.
Note that I did NOT state what is needed. I simply clarified the origin of the question. I even stated that feature/function comarison is moot (and noted that index on expression (which DB2 does NOT support) is a possible match).
Do I need to post incognito to not be shot on sight?

Aside Joe Celko reached beyond orbital speed the moment sequences and identity entered the stage. Little do I care.

Cheers
Serge

-- 
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab

Received on Sun Feb 08 2004 - 17:59:35 CST