diff options
author | Marcelo Vanzin <vanzin@cloudera.com> | 2016-04-25 10:20:51 -0700 |
---|---|---|
committer | Marcelo Vanzin <vanzin@cloudera.com> | 2016-04-25 10:20:51 -0700 |
commit | a680562a6f87a03a00f71bad1c424267ae75c641 (patch) | |
tree | fe08025da6437124c283e9450dd3bbc0e99c411d /python/pyspark/ml/util.py | |
parent | bfda09991398ce44be91997252cf8e5ddd361737 (diff) | |
download | spark-a680562a6f87a03a00f71bad1c424267ae75c641.tar.gz spark-a680562a6f87a03a00f71bad1c424267ae75c641.tar.bz2 spark-a680562a6f87a03a00f71bad1c424267ae75c641.zip |
[SPARK-14744][EXAMPLES] Clean up examples packaging, remove outdated examples.
First, make all dependencies in the examples module provided, and explicitly
list a couple of ones that somehow are promoted to compile by maven. This
means that to run streaming examples, the streaming connector package needs
to be provided to run-examples using --packages or --jars, just like regular
apps.
Also, remove a couple of outdated examples. HBase has had Spark bindings for
a while and is even including them in the HBase distribution in the next
version, making the examples obsolete. The same applies to Cassandra, which
seems to have a proper Spark binding library already.
I just tested the build, which passes, and ran SparkPi. The examples jars
directory now has only two jars:
```
$ ls -1 examples/target/scala-2.11/jars/
scopt_2.11-3.3.0.jar
spark-examples_2.11-2.0.0-SNAPSHOT.jar
```
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes #12544 from vanzin/SPARK-14744.
Diffstat (limited to 'python/pyspark/ml/util.py')
0 files changed, 0 insertions, 0 deletions