Item compression in the Gist index

On Mon, 2022-05-02 at 14:48 +0200, Baca Radim wrote:

we are developing an extension for multidimensional data. We have
created a Gist index that is heavily inspired by a cube extension. Right
now we would like to add some item compression since data items in a
node share a significant portion of a tuple prefix. However, I have no
idea how to handle information stored on a node level in Gists'
compression method. Is there any example? Unfortunately, the cube
extension does not implement item compression.
To be more specific we would like to store in the node a common prefix
for all tuples in the node.
Thanks for any advice,

Perhaps the PostGIS source will inspire you. They are compressing an
entry to its bounding box.

Yours,
Laurenz Albe

On Mon, May 2, 2022 at 02:48:27PM +0200, Baca Radim wrote:

Hi,
we are developing an extension for multidimensional data. We have created a
Gist index that is heavily inspired by a cube extension. Right now we would
like to add some item compression since data items in a node share a
significant portion of a tuple prefix. However, I have no idea how to handle
information stored on a node level in Gists' compression method. Is there
any example? Unfortunately, the cube extension does not implement item
compression.
To be more specific we would like to store in the node a common prefix for
all tuples in the node.

Uh, SP-GiST does prefix compression.

--
Bruce Momjian <bruce@momjian.us> https://momjian.us
EDB https://enterprisedb.com

Indecision is a decision. Inaction is an action. Mark Batterson