[jira] [Resolved] (OPTIQ-410) Allow lattice tiles to satisfy a query by rolling up

["Julian Hyde (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/OPTIQ-410?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Julian Hyde resolved OPTIQ-410.
-------------------------------
       Resolution: Fixed
    Fix Version/s: 0.9.1-incubating

Fixed in http://git-wip-us.apache.org/repos/asf/incubator-optiq/commit/7938e10d.

> Allow lattice tiles to satisfy a query by rolling up
> ----------------------------------------------------
>
>                 Key: OPTIQ-410
>                 URL: https://issues.apache.org/jira/browse/OPTIQ-410
>             Project: Optiq
>          Issue Type: Bug
>            Reporter: Julian Hyde
>            Assignee: Julian Hyde
>             Fix For: 0.9.1-incubating
>
>
> If a lattice has a materialized tile on say columns (x, y, z), allow that tile to be used to evaluate a query on say ( x ) or (y, z). Implementation would require an extra aggregation to roll up from (x, y, z) to the actual column set (y, z).
> Measures would need to be present in the materialization, and be able to be rolled up; for instance, count(a) is rolled up from a partial result p using sum(p). Min, max and sum roll up using themselves.
> You can also compute aggregate functions on grouping columns, for example min( x ), and expressions of aggregate functions.
> Given the a lattice with a single initially materialized tile,
> {code:javascript}
> {
>   auto: false,
>   defaultMeasures: [ {
>     agg: 'count'
>   } ],
>   tiles: [ {
>     dimensions: [ 'the_year', ['t', 'quarter'] ],
>     measures: [ {
>       agg: 'sum',
>       args: 'unit_sales'
>     }, {
>       agg: 'sum',
>       args: 'store_sales'
>     }, {
>       agg: 'count'
>     } ]
>   } ]
> }
> {code}
> Optiq can satisfy the following query from that tile:
> {code:sql}
> select t."the_year",
>   count(*) as c,
>   min("quarter") as q,
>   sum("unit_sales") * 10 as us
> from "foodmart"."sales_fact_1997" as s
> join "foodmart"."time_by_day" as t using ("time_id")
> group by t."the_year"
> {code}
> Note that it rolls up from (the_year, quarter) to (the_year), rolls up count( * ) using sum, rolls up sum(unit_sales), and computes min(quarter) from a column (quarter) that is a dimension, not a measure, of the tile. And it automatically matches the tables joined in the query to the 5 joined tables in the SQL definition of the lattice (not shown above).



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)