Github user aalexandrov commented on the pull request:
https://github.com/apache/flink/pull/880#issuecomment-118035911
@mjsax exactly. Let's at the moment I land in the following situtaion.
1. I want to create a job which includes the [statistics collection
facility proposed by
@tammymendt](https://issues.apache.org/jira/browse/FLINK-1297).
1. In order to do that, my project needs to set `flink-contrib` with scope
`compile` and use it in the final artifact (even if I am building a *thin* jar
which is going to be submitted to a standalone flink cluster).
1. The jar that turns out to be ~ 64 MB. Out of that, half of the bits are
taken by code is *guaranteed* to be in the classpath at rintime by virtue of
being included in the `$FLINK_DIR/lib/flink-dist-$VERSION.jar`.
1. Setting all transitive `flink-*` dependencies via `flink-contrib` as
`provided` reduces the size to ~ 32 MB by default.
---
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.
---
