The drawback is the performance. Kylin is pre-calculation engine, which need define the dimensions and measure before queries. In your case, the count distinct statement will be evaluated at the query running time, not from the pre-calculated cuboid directly .
2017-09-01 23:41 GMT+08:00 Yuxiang Mai <[email protected]>: > Thanks.. > > We will try it. > > BTW, any drawbacks if we directly use in a where condition? > > > > On Fri, Sep 1, 2017 at 11:37 PM, Billy Liu <[email protected]> wrote: > >> The suggested way is to define new column for the if statement, for >> example, in a Hive view. >> >> 2017-09-01 23:08 GMT+08:00 Yuxiang Mai <[email protected]>: >> >>> Hi, experts >>> >>> We are using KYLIN for a Hive table with 3 columns with binary values: 1 >>> for true, 0 for false. For example: is_new, is_recommend, is_discount. >>> We wonder if KYLIN the select if .. from ... like the following: >>> >>> hive> select shop, count(distinct if(is_new =1, item_id, NULL)) from >>> table where dt='xxx' group by shop; >>> >>> or are we mandatory to use where condition is_new=1 ? >>> >>> >>> Thanks for your reply. >>> >>> >>> -- >>> Yuxiang Mai >>> >>> >> > > > -- > Yuxiang Mai > Sun Yat-Sen Unitversity > State Key Lab of Optoelectronic > Materials and Technologies >
