lens-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tao Yan <t...@linkedin.com>
Subject Re: Lens Query Conversion
Date Fri, 05 Aug 2016 20:56:20 GMT
BTW, the query end up with empty result other than the following result in
the 20 minutes demo:

-----------my query result------------
customer_city.name    _c1
Result available in memory, attaching here:

0 rows processed in (57) seconds.


------------demo query result-----------
customer_city.name    _c1

Bangalore    0.0

Hyderabad    2.0

2 rows process in (8) seconds.

On Fri, Aug 5, 2016 at 1:37 PM, Tao Yan <tyan@linkedin.com> wrote:

> Hi Lens Developers,
>
> I am running the example queries, and found the following Lens query:
>
> query execute cube select customer_city_name, store_cost from sales where
> time_range_in(order_time, '2015-04-13-03', '2015-04-13-04')
>
> Was converted to the following hive Query:
>
> SELECT ( customer_city . name ), sum(( sales . store_cost )) FROM
> local_sales_aggr_fact2 sales join local_city_table customer_city on
> sales.customer_city_id = customer_city.id and (customer_city.dt =
> 'latest') WHERE ((((*(( sales . dt ) =  '2015-04-13-04' ) or (( sales .
> dt ) =  '2015-04-13-05' ))* and ((( *sales . order_time* ) >=
>  '2015-04-13 03:00:00' ) and (( *sales . order_time* ) <  '2015-04-13
> 04:00:00' ))))) GROUP BY ( customer_city . name )
>
> The property shows (! grep -A 33333333 <x_ examples/resources/sales-cube.
> xml):
>   <properties>
>     <property name="cube.sales.timed.dimensions.list"
> value="order_time,delivery_time" />
>     *<property name="cube.timedim.partition.order_time" value="ot" />*
>     <property name="cube.timedim.partition.delivery_time" value="dt" />
>     *<property name="cube.timedim.relation.order_time"
> value="delivery_time+[-20 days,-1 hour]" />*
>     <!-- means dt-20days <= ot <= dt-1hour -->
>   </properties>
>
>
> Which means that "order time can be between delivery time -20 days to
> delivery time - 1 hour" according to the 20 minutes demo.
>
> However, the Hive query's 'where' condition use deliver_time equal '
> *2015-04-13-04*' or '*2015-04-13-05*', and how does it find the column
> 'sales.order_time'? Shouldn't it match order_time with partition 'ot'?
>
> Thanks,
> --
>
> *Tao Yan*
> Software Engineer
> Data Analytics Infrastructure Tools and Services
>
>
>
> 206.250.5345
> tyan@linkedin.com
> https://www.linkedin.com/in/taousc
>



-- 

*Tao Yan*
Software Engineer
Data Analytics Infrastructure Tools and Services



206.250.5345
tyan@linkedin.com
https://www.linkedin.com/in/taousc

Mime
View raw message