Dear Jay,

1 - a) To use Flink Streaming all you need to do is to have it as a dependency as described in the documentation. [1] This would build it against the last snapshot that we have deployed which is updated on every successful continuous integration build (typically multiple times a day). If you want to still build Flink locally 'mvn clean install -DskipTests' does the trick, then you can use it as a dependency in your project.

1 - b) The purpose of staging is to have a repository for modules that the Flink community marks as experimental before moving them top level as you have suggested.

2) The one and only Flink Streaming is currently located under staging, but you can expect it to be moved top level in the very near future. As you have seen in 1a) this system does not affect users directly, but is rather an internal development policy. 

[1] https://ci.apache.org/projects/flink/flink-docs-master/apis/streaming_guide.html

On Sun, Aug 9, 2015 at 6:32 PM, jay vyas <jayunit100.apache@gmail.com> wrote:
Hi flink.  

I've been attempting to build a flink-streaming application using the 0.10-SNAPSHOT from head.

I notice that flink-streaming is hidden under "flink-staging".... 

1) What is the right way to build a flink-streaming app against head, and what is the purpose of the flink-staging directory? 

2) Is there is a more stable flink streaming api, such that staging/flink-streaming is an experimental module?  

(context = We are building the BigPetStore-Flink reference implementation... so we'd like to do this idiomatically and I'm happy to use the flink-staging/ work if thats the right way to move forward.)

Thanks !

--
jay vyas