Github user stephenh commented on the pull request:
https://github.com/apache/spark/pull/4166#issuecomment-72877246
@MattWhelan I believe you that loadClass is preferable to findClass, but
can you articulate why? And maybe as a comment on this SO post so that future
people that copy/paste from the 1st answer can know as well? :-)
http://stackoverflow.com/questions/5445511/how-do-i-create-a-parent-last-child-first-classloader-in-java-or-how-to-overr
Tangentially, I'd eventually like to port/copy/paste the Hadoop
ApplicationClassLoader (which I believe they copy/pasted from Jetty) so that we
can ignore any org.apache.spark.* classes that might accidentally be in the
user's jar:
https://github.com/apache/hadoop/blob/trunk/hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/util/ApplicationClassLoader.java
I'm also half-tempted to copy/paste their ApplicationClassLoader into a
separate utility project, so that going forward Jetty/Hadoop/Spark/etc. could
all reuse the same tested functionality instead of copy/pasting the code
around. But then I'd have to publish to maven central, it'd be another
dependency to worry about, etc., etc...
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
