[
https://issues.apache.org/jira/browse/TWILL-63?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14947772#comment-14947772
]
ASF GitHub Bot commented on TWILL-63:
-------------------------------------
Github user chtyim commented on a diff in the pull request:
https://github.com/apache/incubator-twill/pull/66#discussion_r41459425
--- Diff:
twill-core/src/main/java/org/apache/twill/internal/ApplicationBundler.java ---
@@ -178,8 +181,42 @@ public void createBundle(Location target,
Iterable<Class<?>> classes, Iterable<U
}
}
+ /**
+ * Creates a {@link ByteArrayOutputStream} which includes all the given
classes and
+ * all the classes that they depended on.
+ * The {@link ByteArrayOutputStream}
+ * will also include all classes and resources under the packages as
given as include packages
+ * in the constructor.
+ *
+ * @param resources Extra resources to put into the jar file. If
resource is a jar file, it'll be put under
+ * lib/ entry, otherwise under the resources/ entry.
+ * @param classes Set of classes to start the dependency traversal.
+ * @return ByteArrayOutputStream
+ * @throws IOException
+ */
+ public ByteArrayOutputStream getBundleAsStream(Iterable<Class<?>>
classes,
--- End diff --
So this could end up using a lot of heap space (and double it when
`ByteArrayOutputStream.toByteArray()` is being called). Why it needs to be kept
all in memory?
> Speed up application launch time
> --------------------------------
>
> Key: TWILL-63
> URL: https://issues.apache.org/jira/browse/TWILL-63
> Project: Apache Twill
> Issue Type: Improvement
> Components: yarn
> Affects Versions: 0.2.0-incubating
> Reporter: Terence Yim
> Assignee: Shankar Selvam
> Fix For: 0.7.0-incubating
>
>
> Currently when launching an application, two new jars are always created
> locally, one for AM (appMaster.jar) and one for Container (container.jar) and
> copied to HDFS before submitting the application. The jar files could
> potentially be big and if it doesn't changed, it should require copying to
> HDFS again.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
