drill-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jason Altekruse" <altekruseja...@gmail.com>
Subject Re: Review Request 31871: 2406 - part 2 - (subset of 2060 part 2) enable interpreted expression evaluation at planning time
Date Mon, 09 Mar 2015 23:12:45 GMT


> On March 9, 2015, 11:08 p.m., Jason Altekruse wrote:
> > exec/java-exec/src/main/java/org/apache/drill/exec/planner/sql/DrillSqlOperator.java,
line 54
> > <https://reviews.apache.org/r/31871/diff/1/?file=889812#file889812line54>
> >
> >     The reason this is kept in a variable called random is that we have been calling
this probperty isRandom in our function definitions. Calcite uses the opposite flag, of deterministic
to refer to the same property. I decided to push this all the way to the point where we connect
to calicte to try to avoid confusion or major refactoring for little benefit.

property*


- Jason


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/31871/#review75807
-----------------------------------------------------------


On March 9, 2015, 10:07 p.m., Jason Altekruse wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/31871/
> -----------------------------------------------------------
> 
> (Updated March 9, 2015, 10:07 p.m.)
> 
> 
> Review request for drill and Jacques Nadeau.
> 
> 
> Bugs: DRILL-2406
>     https://issues.apache.org/jira/browse/DRILL-2406
> 
> 
> Repository: drill-git
> 
> 
> Description
> -------
> 
> To enable planning rules to take advantage of constant expressions in queries, or to
allow them to run queries against small datasets, such as partition information we need to
e able to evaluate expressions at planning time. Expression evaluation in the regular expecuation
path is relatively expensive, as it invovles java code generation, compilation and JITing
expression trees. An interpreted expression system was added recently to allow evaluating
an expression without generating java code at runtime. This patch exposes that work (after
some refactorings in 2143 and the first part of 2406) to the planning context for use in optimizer
rules.
> 
> 
> Diffs
> -----
> 
>   exec/java-exec/src/main/java/org/apache/drill/exec/expr/fn/DrillFunctionRegistry.java
00aaec6 
>   exec/java-exec/src/main/java/org/apache/drill/exec/memory/TopLevelAllocator.java 2a28bcb

>   exec/java-exec/src/main/java/org/apache/drill/exec/ops/BufferManager.java PRE-CREATION

>   exec/java-exec/src/main/java/org/apache/drill/exec/ops/FragmentContext.java aa1dffd

>   exec/java-exec/src/main/java/org/apache/drill/exec/ops/QueryContext.java c881432 
>   exec/java-exec/src/main/java/org/apache/drill/exec/planner/logical/DrillReduceAggregatesRule.java
93fff35 
>   exec/java-exec/src/main/java/org/apache/drill/exec/planner/physical/visitor/InsertLocalExchangeVisitor.java
907fcb1 
>   exec/java-exec/src/main/java/org/apache/drill/exec/planner/sql/DrillSqlOperator.java
6b54c43 
>   exec/java-exec/src/main/java/org/apache/drill/exec/work/foreman/Foreman.java 409450f

> 
> Diff: https://reviews.apache.org/r/31871/diff/
> 
> 
> Testing
> -------
> 
> Cluster tests completed, waiting on unit tests.
> 
> 
> Thanks,
> 
> Jason Altekruse
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message