[
https://issues.apache.org/jira/browse/FLINK-2458?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14661751#comment-14661751
]
ASF GitHub Bot commented on FLINK-2458:
---------------------------------------
Github user sachingoel0101 commented on a diff in the pull request:
https://github.com/apache/flink/pull/970#discussion_r36513272
--- Diff:
flink-core/src/main/java/org/apache/flink/api/common/operators/CollectionExecutor.java
---
@@ -501,4 +536,22 @@ public int getSuperstepNumber() {
return (T) previousAggregates.get(name);
}
}
+
+ private static final class DoingNothing implements Callable<Path>{
+ private Path entry;
+
+ public DoingNothing(Path entry){
+ this.entry = entry;
+ }
+
+ @Override
+ public Path call() throws IOException{
+ try{
+ LocalFileSystem fs = (LocalFileSystem)
entry.getFileSystem();
+ return entry.isAbsolute() ? new
Path(entry.toUri().getPath()): new Path(fs.getWorkingDirectory(),entry);
+ } catch (ClassCastException e){
+ throw new RuntimeException("Collection
execution must have only local file paths");
--- End diff --
Yeah. That would make more sense.
> Distributed Cache doesn't work with iterations.
> -----------------------------------------------
>
> Key: FLINK-2458
> URL: https://issues.apache.org/jira/browse/FLINK-2458
> Project: Flink
> Issue Type: Bug
> Reporter: Sachin Goel
> Assignee: Sachin Goel
>
> Distributed cache files cannot be accessed when working in Iterative
> contexts. While constructing the context in {{AbstractIterativePactTask}},
> the distributed entries are not passed on to runtime context.
> My commit will add this and also unify the distributed cache system to work
> everywhere, including in the Collection Environment.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
