Pgsql: Support Grouping Sets, Cube, and Rollup | Hacker News Reader

Pgsql: Support Grouping Sets, Cube, and Rollup

postgresql.org

69 pointsamitlan20 comments

I was just looking at the wiki page https://wiki.postgresql.org/wiki/Grouping_Sets yesterday hoping that one day it would be in.

I now see that postgresql is the one true database. Everyone should abandon all other databases that don't have special features (ex: sqlite embedded).

> Everyone should abandon all other databases that don't have special features (ex: sqlite embedded).

SQLite embedded does have special features. It's embedded.

I believe GP used that as an example of what not to abandon

Is "ex" short for "example" or "except"?

I use it for both. In that case it was 'example'.

Just added a line to that page saying it's in, :)

Wish they'd give some example sql statements to fully appreciate it.

Maybe ping depesz (http://www.depesz.com/) about it. He has that great "waiting for postgres X.X" column at his blog, where he illustrates new things.

It's on the developer version docs: http://www.postgresql.org/docs/devel/static/queries-table-ex...

Also here: https://wiki.postgresql.org/wiki/Grouping_Sets

It's a commit message, not the docs though ;).

PGSQL just keeps getting better! Even gives Oracle a run for its money in a lot of areas.

I am a huge fan and advocate of PG but just for some perspective: CUBE and ROLLUP were present in SQL Server in the '90s.

I agree with you about PostgreSQL being late at adding these features, but SQL Server added CUBE and ROLLUP in 2008. They had GROUPING SETS longer, I think since 2000.

Here's an article on them from '99. Version 6.5 was released in '96.

http://sqlmag.com/database-development/sql-server-65-cube-ro...

So does this mean I could use pg instead is SSAS for cube analysis?

You could before. The benefit of OLAP databases and other analytical engines is only partly due to the focus of the query languages. Storage format, and calculation engine requirements are vastly different between OLTP and OLAP.

While the constructs in the linked article increase the expressive power of SQL, they are more icing than cake for an OLAP load.

SSAS and other cube (MOLAP) tools create persistent cubes as materialized views. Databases that do ROLAP processing create "cubes on the fly" which persist for the duration of the query.

If PG adopts a materialized view that supports controlled automatic maintenance, then you'd have something similar to what the MOLAP tools do.

Note that MOLAP is a trap of sorts (no pun intended).

Apart from the complexity of cube updating, why do you say that?

What makes MOLAP a trap? I'm only generally acquainted with cube analysis.

Reply on news.ycombinator.com