Thread: B-tree leaf node structure

B-tree leaf node structure

From
edwardyf
Date:
If the index is on an attribute with duplicate values. will it be:
1) one index tuple for each row, though with the same value, or
2) one index tuple for each value, containing a list of row ids.

thanx
-- 
View this message in context: http://www.nabble.com/B-tree-leaf-node-structure-tp26004939p26004939.html
Sent from the PostgreSQL - hackers mailing list archive at Nabble.com.



Re: B-tree leaf node structure

From
Tom Lane
Date:
edwardyf <edwardyf@gmail.com> writes:
> If the index is on an attribute with duplicate values. will it be:
> 1) one index tuple for each row, though with the same value, or
> 2) one index tuple for each value, containing a list of row ids.

1.
        regards, tom lane


Re: B-tree leaf node structure

From
Jeff Davis
Date:
On Wed, 2009-10-21 at 23:55 -0700, edwardyf wrote:
> If the index is on an attribute with duplicate values. will it be:
> 1) one index tuple for each row, though with the same value, or
> 2) one index tuple for each value, containing a list of row ids.

As Tom already pointed out, #1 is the answer.

However, I'd like to add that there's a feature that never quite made it
called Grouped Index Tuples (GIT) that might still be viable:

http://community.enterprisedb.com/git/

Regards,Jeff Davis