Mailing-List: contact dev-help@hive.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hive.apache.org
Content-Type: multipart/alternative;
 boundary="===============8370672460104431602=="
MIME-Version: 1.0
Subject: Re: Review Request 63343: HIVE-17766
From: =?utf-8?q?Jes=C3=BAs_Camacho_Rodr=C3=ADguez?=
 <jcamachorodriguez@hortonworks.com>
To: Ashutosh Chauhan <hashutosh@apache.org>,
 Vineet Garg <vgarg@hortonworks.com>
Cc: hive <dev@hive.apache.org>, =?utf-8?q?Jes=C3=BAs_Camacho_Rodr=C3=ADguez?=
 <jcamachorodriguez@hortonworks.com>
Date: Mon, 30 Oct 2017 18:49:58 -0000
Message-ID: <20171030184958.50427.42012@reviews-vm2.apache.org>
Auto-Submitted: auto-generated
Sender: =?utf-8?q?Jes=C3=BAs_Camacho_Rodr=C3=ADguez_=3Cnoreply=40reviews=2E?=@reviews.apache.org,
	=?utf-8?q?apache=2Eorg=3E?=@reviews.apache.org
References: <20171026230832.51101.75748@reviews-vm2.apache.org>
In-Reply-To: <20171026230832.51101.75748@reviews-vm2.apache.org>
Reply-To: =?utf-8?q?Jes=C3=BAs_Camacho_Rodr=C3=ADguez_=3Cjcamachorodriguez?=@reviews.apache.org,
	=?utf-8?q?=40hortonworks=2Ecom=3E?=@reviews.apache.org
archived-at: Mon, 30 Oct 2017 18:50:03 -0000

--===============8370672460104431602==
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 8bit


> On Oct. 26, 2017, 11:08 p.m., Ashutosh Chauhan wrote:
> > ql/src/java/org/apache/hadoop/hive/ql/parse/CalcitePlanner.java
> > Lines 2136-2137 (original), 2119-2120 (patched)
> > <https://reviews.apache.org/r/63343/diff/2/?file=1869813#file1869813line2136>
> >
> >     Shall we try to get equi joins keys and populate them in operator here? Will it happen later in planning.
> >     Either way good to add a comment about this.
> 
> Jesús Camacho Rodríguez wrote:
>     This is in Calcite operators, not in Hive. The reason keys are there is because semijoin extends equijoin. I do not know if they are effectively used, the logic here was pushing the computation of expressions to children project, but that seems to be happening when we trigger rewriting rules in any case, thus I do not think we need to do it here (maybe this code was written before some of the rules that take care of this were enabled in Hive?).

I discovered some tests in previous iteration of the patch where filters that were applying to only one input where not pushed below the semijoin.
I have fixed this logic to push whatever we can push below the semijoin and create the semijoin condition with the rest.


- Jesús


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/63343/#review189370
-----------------------------------------------------------


On Oct. 28, 2017, 8:06 p.m., Jesús Camacho Rodríguez wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/63343/
> -----------------------------------------------------------
> 
> (Updated Oct. 28, 2017, 8:06 p.m.)
> 
> 
> Review request for hive, Ashutosh Chauhan and Vineet Garg.
> 
> 
> Bugs: HIVE-17766
>     https://issues.apache.org/jira/browse/HIVE-17766
> 
> 
> Repository: hive-git
> 
> 
> Description
> -------
> 
> HIVE-17766
> 
> 
> Diffs
> -----
> 
>   itests/src/test/resources/testconfiguration.properties a4648beae1ec18c88bab35ae3e7f9d9f202e40e5 
>   ql/src/java/org/apache/hadoop/hive/ql/exec/CommonJoinOperator.java 3573d07e9a3bab4dabdc0b73292bd02ad05cbc89 
>   ql/src/java/org/apache/hadoop/hive/ql/parse/CalcitePlanner.java 46493ac006d00f2e6c0054da2438c0b01c1e3fd3 
>   ql/src/java/org/apache/hadoop/hive/ql/parse/SemanticAnalyzer.java 5ccb69ac4824a2002b2bddb1ad206cb7acd3ace3 
>   ql/src/java/org/apache/hadoop/hive/ql/parse/TypeCheckProcFactory.java 4ddf1d5b5e87a43f7f606ce3c09ac295f5319333 
>   ql/src/test/queries/clientpositive/semijoin6.q PRE-CREATION 
>   ql/src/test/results/clientpositive/llap/semijoin6.q.out PRE-CREATION 
> 
> 
> Diff: https://reviews.apache.org/r/63343/diff/5/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Jesús Camacho Rodríguez
> 
>


--===============8370672460104431602==--