Hi, ExecutionEnvironment has a static field contextEnvironment. When you call getExecutionEnvironment it checks whether this is null. If it is null it will return a LocalEnvironment, otherwise it will return the context environment.
In Client.run() the Client creates a ContextEnvironment and calls setAsContext() on it. This will put the env into the contextEnvironment field of ExecutionEnvironment. I think you don't have to deal with all this though. Can you just wrap a normal ExecutionEnvironment inside your streaming-specific environment? In that case, you could just use ExecutionEnvironemnt.getExecutionEnvironment and wrap the result of that. This is how it works in the Scala API. Have a look at ExecutionEnvironment.scala, it's really just a wrapper. Hope that helps. :D Cheers, Aljoscha On Fri, Nov 7, 2014 at 3:24 PM, Gyula Fóra <[email protected]> wrote: > Hey Guys, > > The .getExecutionEnvironment() method for the StreamingexecutionEnvironment > does not work properly because we always return LocalEnvironment for > running on the minicluster. > > I was trying to figure out how to fix this but I got lost in the code > trying to find how you determine whether the program was executed > standalone or the command line client. > > Could you help me out with some pointers here? > > Regards, > Gyula
