Skip to content

Spark 615 map partitions with index callable from java #16

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
holdenk wants to merge 25 commits into from
Closed

Spark 615 map partitions with index callable from java #16

Changes from 1 commit
Commits
Show all changes
25 commits
Select commit Hold shift + click to select a range
215a9bf
Fix Java API for mapPartitionsWithIndex
holdenk Feb 17, 2014
e2331ed
Check all the values
holdenk Feb 17, 2014
0d624bf
Add missing class
holdenk Feb 24, 2014
4421ecc
Use fakeClassTag
holdenk Feb 27, 2014
8d849a1
Fix Java API for mapPartitionsWithIndex
holdenk Feb 17, 2014
958efa4
Check all the values
holdenk Feb 17, 2014
6ad1a3c
Add missing class
holdenk Feb 24, 2014
e64e1ad
Use fakeClassTag
holdenk Feb 27, 2014
8bfd3f3
It compiles with the Java 8 happy pandas
holdenk Mar 8, 2014
b6a613f
merge
holdenk Mar 8, 2014
36c7831
Remove old function
holdenk Mar 8, 2014
79d1bc1
Check all the values
holdenk Feb 17, 2014
ec80d7a
Add missing class
holdenk Feb 24, 2014
e4962ab
Use fakeClassTag
holdenk Feb 27, 2014
f484afc
Fix Java API for mapPartitionsWithIndex
holdenk Feb 17, 2014
4eb9c0f
Add missing class
holdenk Feb 24, 2014
96a86c7
It compiles with the Java 8 happy pandas
holdenk Mar 8, 2014
df6922a
Remove old function
holdenk Mar 8, 2014
4fa42b0
Revert "Remove old function (accidently removed a file)"
holdenk Mar 11, 2014
2aa2fc9
Make it work with the new Java API style
holdenk Mar 11, 2014
cc4050d
Add Java 8 API test
holdenk Mar 15, 2014
ee6ebea
Some CR feedback
holdenk Apr 8, 2014
780120d
Remove MappartitionsWithIndeFunction
holdenk Apr 8, 2014
a5cd2fa
Fix Java API
holdenk Apr 8, 2014
b231375
Fix the classtag bits
holdenk Apr 8, 2014
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
Fix Java API for mapPartitionsWithIndex
  • Loading branch information
@holdenk
holdenk committed Mar 11, 2014
commit f484afce126108375e1205f2b68f1577c4ef9598
12 changes: 7 additions & 5 deletions core/src/main/scala/org/apache/spark/api/java/JavaRDDLike.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,11 +72,13 @@ trait JavaRDDLike[T, This <: JavaRDDLike[T, This]] extends Serializable {
* Return a new RDD by applying a function to each partition of this RDD, while tracking the index
* of the original partition.
*/
def mapPartitionsWithIndex[R: ClassTag](
f: JFunction2[java.lang.Integer, java.util.Iterator[T], java.util.Iterator[R]],
preservesPartitioning: Boolean = false): JavaRDD[R] =
new JavaRDD(rdd.mapPartitionsWithIndex(((a,b) => f(a,asJavaIterator(b))),
preservesPartitioning))
def mapPartitionsWithIndex[R](f: MapPartitionsWithIndexFunction[T, R],
Copy link
Contributor

@pwendell pwendell Mar 16, 2014

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just wondering - why isn't this version callable from Java? It seems better to me not to define a new Function class... I'm trying to understand why the existing one can't be called as-is (modulo the classtag stuff).

Copy link
Contributor Author

@holdenk holdenk Mar 17, 2014

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I believe it was just the classtag stuff. The new function class was for convience and in keeping with the other functions being defined (FlatMapFunction etc.). I can get of the new function class if you want.

Copy link
Contributor

@pwendell pwendell Mar 17, 2014

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

hey @holdenk now that I look more I'm all confused because it looks like FlatMapFunction is e.g. used for mapPartitions despite it's name. I'll have to defer to @mateiz on this one who has more recently looked at the java API.

Copy link
Contributor

@mateiz mateiz Mar 17, 2014

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah I agree, did you try using the old one? It lacked a test but maybe it can work. Throughout the Java API we only pass the ClassTag for java.lang.Object, we don't try to pass a real class tag.

Copy link
Contributor Author

@holdenk holdenk Mar 25, 2014

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I tried calling the old one, but that was pre the big Java API refactoring, and it didn't work.

Copy link
Contributor

@mateiz mateiz Apr 5, 2014

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Try updating this to just take a Function2<Integer, java.util.Iterator<A>, java.util.Iterator<B>>. I'm pretty sure it would work and we won't need a new type of class.

preservesPartitioning: Boolean = false): JavaRDD[R] = {
import scala.collection.JavaConverters._
def fn = (a: Int, b: Iterator[T]) => f.apply(a, asJavaIterator(b)).asScala
val newRdd = rdd.mapPartitionsWithIndex(fn, preservesPartitioning)(f.elementType())
new JavaRDD(newRdd)(f.elementType())
}

/**
* Return a new RDD by applying a function to all elements of this RDD.
Expand Down