Hi Roman,
There are some potential advantages.
Let's still use this as example.
SELECT * FROM foo WHERE a > 100 or b < 1000;
Both column a and b have B-Tree indexes, note that it doesn't have to be bitmap
index.
Bitmap Table Scan on foo
-> BitmapOr
-> Bitmap Index Scan on
Hello,
Since I was among the people who raised some concerns about this during the
vote I would like to mention that it was mostly about the fact it came very
late in the release cycle.
It is better to avoid breaking changes in the RC phase since it may be
difficult for people to react or test
I've seen the commit coming through and I do like the approach of keeping
the execution logic separate, thanks Julian!
Best,
Stamatis
On Mon, Jun 1, 2020 at 9:02 PM Julian Hyde wrote:
> Just to finish this thread... I've now merged 3946.
>
> I did what Stamatis suggested, and moved the DDL
I'm pleased there are a variety of approaches. People should use
whichever works for them and their use case.
The so-called "rule-based" approach is definitely useful for
OLTP-style queries where you are accessing a few rows and you need to
plan quickly.
Haisheng mentioned bitmap indexes a while
Just to finish this thread... I've now merged 3946.
I did what Stamatis suggested, and moved the DDL classes into core.
However, I kept the execution logic (e.g. creating the table when
someone sends a SqlCreateTable command) in server. To do this, I
created a new interface DdlExecutor, and moved
Hi Xiening,
the example was synthetic. What is still not clear for me is how to
exploit index sortedness with a rule based approach. As far as I
understand with this approach we need to write complex rules (for
example [1]) that should decide which index is useful and which is not
useful. These
Hi Roman,
The example you mentioned is an advanced scenario. Note that there are
different types of index, such as clustered index, secondary index, covered and
non-covered index. In your case, typical OLTP/OLAP optimizer would create an
index-based join on top of the range table scan (or
Sylvain Crozon created CALCITE-4037:
---
Summary: AggregateProjectMergeRule doesn't always respect column
aliases
Key: CALCITE-4037
URL: https://issues.apache.org/jira/browse/CALCITE-4037
Project: