I have a optimizer question I was hoping for someone well versed could
provide some input to:
As part of my work on DERBY-2998, the implementation of the ROW_NUMBER()
window function, I'm having trouble with the optimizer and its (lack of)
pushing predicates down the query tree
Army,
Thanks for taking the time to have a look at this.
To add to the info I see the same kind of queryplan in the similar
query where the aliased expression is exchanged with 'a*b':
SELECT * FROM (SELECT a*b AS R, T.* FROM T) AS TR WHERE R 3;
So I'm starting to wonder if this is a general
Bryan Pendleton wrote:
As part of my work on DERBY-2998, the implementation of the
ROW_NUMBER() window function, I'm having trouble with the optimizer
and its (lack of) pushing predicates down the query tree.
Hi Thomas Army, this is a very interesting discussion and I'm learning
a lot by
As part of my work on DERBY-2998, the implementation of the
ROW_NUMBER() window function, I'm having trouble with the optimizer
and its (lack of) pushing predicates down the query tree.
Hi Thomas Army, this is a very interesting discussion and I'm learning
a lot by following it, so please
Army wrote:
Oops, sorry about that. I was just reading your email, I didn't read
the comments on DERBY-2998. Apologies for the noise...
I wasn't expecting you to either - apologies if I sounded like that. And
it's by no means noise to me - it was acutally truly rewarding to see
that my
Thomas Nielsen wrote:
If you look at my comments in DERBY-2998, I acutally got this far a
while ago.
Oops, sorry about that. I was just reading your email, I didn't read
the comments on DERBY-2998. Apologies for the noise...
One of the results of adding the extra if
Bryan Pendleton wrote:
[snip]
With the new restriction approach that you are proposing for DERBY-2998,
am I correct that you are suggesting we might invent a new ResultSet
node implementation, which computes the ROW_NUMBER() value for its rows?
Almost.
In my patch I'm suggesting to
Thomas Nielsen wrote:
As part of my work on DERBY-2998, the implementation of the ROW_NUMBER()
window function, I'm having trouble with the optimizer and its (lack of)
pushing predicates down the query tree.
[ snip ]
To add to the info I see the same kind of queryplan in the similar query
Hi Optimizer Enthusiasts!
I've been thinking about predicate pushdown and have a question. The question
specifically involves this section of code in OptimizerImpl:
double newCost = currentCost.getEstimatedCost();
double pullCost = 0.0;
Bryan Pendleton wrote:
I have two questions about this code, both related to the edge-cases of
floating point behavior:
I think both of your questions fall into the category of DERBY-1260, which was
filed for investigation of how infinite cost estimates affect Derby costing
calculations. I
I think both of your questions fall into the category of DERBY-1260,
I agree. I will attach these thoughts to that issue.
Thank you very much for the thoughtful reply! There is much to learn
about the optimizer.
bryan
11 matches
Mail list logo