Mailing-List: contact dev-help@kylin.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@kylin.apache.org
MIME-Version: 1.0
In-Reply-To: <tencent_27523D003EBC01215CFA22CB@qq.com>
References: <tencent_27523D003EBC01215CFA22CB@qq.com>
Date: Tue, 12 Jan 2016 17:29:54 +0800
Message-ID: 
 <CAHRce1NQBLpCqrHUBK8JJvcHFdThUkEZEhb-GHMxcFBuo2V7_A@mail.gmail.com>
Subject: Re: A problem,maybe a bug, when querying in kylin using not in
From: Li Yang <liyang@apache.org>
To: dev@kylin.apache.org
Content-Type: multipart/alternative; boundary=001a11395e543d51e005291fb239

--001a11395e543d51e005291fb239
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Let's continue discussion in the JIRA, KYLIN-1294.

On Thu, Jan 7, 2016 at 11:40 AM, =E6=9B=BE=E7=BB=8F <1032195826@qq.com> wro=
te:

> In the case you give lstg_format_name does not contain null, I tested in
> my cube, and it's also OK.
> While, when I execute the following sql, where "a" is not a member of
> "KYLIN_TV_DIM_AREA"."TV_AREA_NAME", the "TV_AREA_NAME" of "=E9=BB=98=E8=
=AE=A4=E5=9C=B0=E5=8C=BA" will still
> appear in the result set. I think this is also a related problem in the b=
ug.
> The sql is :
> SELECT "KYLIN_TV_DIM_AREA"."TV_AREA_NAME" AS "TV_AREA_NAME",
>        SUM("KYLIN_VIEW_TVAD_SUMMARY"."INCOME") AS "sum_INCOME_ok",
>        SUM(1) AS "sum_Number_of_Records_ok",
>        SUM("KYLIN_VIEW_TVAD_SUMMARY"."VV") AS "sum_VV_ok"
>   FROM "BD_WAREHOUSE"."KYLIN_VIEW_TVAD_SUMMARY" "KYLIN_VIEW_TVAD_SUMMARY"
>   LEFT JOIN "BD_WAREHOUSE"."KYLIN_TV_DIM_AREA" "KYLIN_TV_DIM_AREA"
>     ON ("KYLIN_VIEW_TVAD_SUMMARY"."AREA" =3D
> "KYLIN_TV_DIM_AREA"."TV_AREA_ID")
>  WHERE (("KYLIN_TV_DIM_AREA"."TV_AREA_NAME" not IN ('=E7=BE=8E=E5=89=A7',=
 'a', '=E9=BB=98=E8=AE=A4=E5=9C=B0=E5=8C=BA')))
>  GROUP BY "KYLIN_TV_DIM_AREA"."TV_AREA_NAME";
>
>
>
>
>
> ------------------ =E5=8E=9F=E5=A7=8B=E9=82=AE=E4=BB=B6 -----------------=
-
> =E5=8F=91=E4=BB=B6=E4=BA=BA: "Li Yang";<liyang@apache.org>;
> =E5=8F=91=E9=80=81=E6=97=B6=E9=97=B4: 2016=E5=B9=B41=E6=9C=887=E6=97=A5(=
=E6=98=9F=E6=9C=9F=E5=9B=9B) =E4=B8=AD=E5=8D=8811:28
> =E6=94=B6=E4=BB=B6=E4=BA=BA: "dev"<dev@kylin.apache.org>;
>
> =E4=B8=BB=E9=A2=98: Re: A problem,maybe a bug, when querying in kylin usi=
ng not in
>
>
>
> Confirm this is a bug. Not sure of the root cause, but I can reproduce on
> 2.x and 1.x branch though they yield different exceptions.
>
> Open a JIRA for further investigation.
> https://issues.apache.org/jira/browse/KYLIN-1294
>
> Meanwhile worth note that this test query did pass with no problem, givin=
g
> ('FP-GTC', 'ABIN') are valid values and lstg_format_name does not contain
> null.
>
> select lstg_format_name, sum(price) as GMV
>  from test_kylin_fact
>  where lstg_format_name not in ('FP-GTC', 'ABIN')
>  group by lstg_format_name
>
>
>
> On Thu, Jan 7, 2016 at 10:54 AM, Julian Hyde <jhyde@apache.org> wrote:
>
> > Yes, it looks as if you have a data correctness issue. If I were you I=
=E2=80=99d
> > look at the plan and see if it makes sense.
> >
> > Your second case, the NullPointerException issue, looks a lot like
> > https://issues.apache.org/jira/browse/CALCITE-980 <
> > https://issues.apache.org/jira/browse/CALCITE-980>.
> >
> > Julian
> >
> > > On Jan 6, 2016, at 6:28 PM, =E6=9B=BE=E7=BB=8F <1032195826@qq.com> wr=
ote:
> > >
> > > Thank you very much for answering.
> > > In my result ,not only null in H_CODE is skipped, but also the lines
> > where H_CODE is 'B' and 'O' are all skipped. I think this.
> > > I think this is not only a prolem of null.
> >
> >
>

--001a11395e543d51e005291fb239--