[ 
https://issues.apache.org/jira/browse/SPARK-9999?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14957926#comment-14957926
 ] 

Michael Armbrust commented on SPARK-9999:
-----------------------------------------

[~sandyr] did you look at the test cases [in 
scala|https://github.com/marmbrus/spark/blob/d0277f5013fd9e5e758c607b5c833cf5aa7bb93c/sql/core/src/test/scala/org/apache/spark/sql/DatasetSuite.scala]
 and 
[java|https://github.com/marmbrus/spark/blob/d0277f5013fd9e5e758c607b5c833cf5aa7bb93c/sql/core/src/test/java/test/org/apache/spark/sql/JavaDatasetSuite.java]
 linked from the attached design doc?

In Scala, users should never have to think about Encoders as long as their data 
can be represented as primitives, case classes, tuples, or collections.  
Implicits (provided by {{sqlContext.implicits._}}) automatically pass the 
required information to the function.  

In Java, the compiler is not helping us out as much, so the user must do as you 
suggest.  The prototype shows {{ProductEncoder.tuple(Long.class, Long.class)}}, 
but we will have a similar interface that works for class objects for POJOs / 
JavaBeans.  The problem with doing this using a registry (like kryo in RDDs 
today) is that then you aren't finding out the object type until you have an 
example object from realizing the computation.  That is often too late to do 
the kinds of optimizations that we are trying to enable.  Instead we'd like to 
statically realize the schema at Dataset construction time.

Encoders are just an encapsulation of the required information and provide an 
interface if we ever want to allow someone to specify a custom encoder.

Regarding the performance concerns with reflection, the implementation that is 
already present in Spark master ([SPARK-10993] and [SPARK-11090]) is based on 
catalyst expressions.  Reflection is done once on the driver, and the existing 
code generation caching framework is taking care of caching generated encoder 
bytecode on the executors.

> RDD-like API on top of Catalyst/DataFrame
> -----------------------------------------
>
>                 Key: SPARK-9999
>                 URL: https://issues.apache.org/jira/browse/SPARK-9999
>             Project: Spark
>          Issue Type: Story
>          Components: SQL
>            Reporter: Reynold Xin
>            Assignee: Michael Armbrust
>
> The RDD API is very flexible, and as a result harder to optimize its 
> execution in some cases. The DataFrame API, on the other hand, is much easier 
> to optimize, but lacks some of the nice perks of the RDD API (e.g. harder to 
> use UDFs, lack of strong types in Scala/Java).
> The goal of Spark Datasets is to provide an API that allows users to easily 
> express transformations on domain objects, while also providing the 
> performance and robustness advantages of the Spark SQL execution engine.
> h2. Requirements
>  - *Fast* - In most cases, the performance of Datasets should be equal to or 
> better than working with RDDs.  Encoders should be as fast or faster than 
> Kryo and Java serialization, and unnecessary conversion should be avoided.
>  - *Typesafe* - Similar to RDDs, objects and functions that operate on those 
> objects should provide compile-time safety where possible.  When converting 
> from data where the schema is not known at compile-time (for example data 
> read from an external source such as JSON), the conversion function should 
> fail-fast if there is a schema mismatch.
>  - *Support for a variety of object models* - Default encoders should be 
> provided for a variety of object models: primitive types, case classes, 
> tuples, POJOs, JavaBeans, etc.  Ideally, objects that follow standard 
> conventions, such as Avro SpecificRecords, should also work out of the box.
>  - *Java Compatible* - Datasets should provide a single API that works in 
> both Scala and Java.  Where possible, shared types like Array will be used in 
> the API.  Where not possible, overloaded functions should be provided for 
> both languages.  Scala concepts, such as ClassTags should not be required in 
> the user-facing API.
>  - *Interoperates with DataFrames* - Users should be able to seamlessly 
> transition between Datasets and DataFrames, without specifying conversion 
> boiler-plate.  When names used in the input schema line-up with fields in the 
> given class, no extra mapping should be necessary.  Libraries like MLlib 
> should not need to provide different interfaces for accepting DataFrames and 
> Datasets as input.
> For a detailed outline of the complete proposed API: 
> [marmbrus/dataset-api|https://github.com/marmbrus/spark/pull/18/files]
> For an initial discussion of the design considerations in this API: [design 
> doc|https://docs.google.com/document/d/1ZVaDqOcLm2-NcS0TElmslHLsEIEwqzt0vBvzpLrV6Ik/edit#]



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org

Reply via email to