The recommend way to bring graphhopper to production is by running the graphhopper.sh start script.
Is there any chance to bring back the possibility to build war archives? That would make it much easier to run graphopper within managed environments like Elastic Beanstalk etc.
As we do not use this deployment method and currently also have no customers nor know other users that use this you would need to convince us somehow and we would need to make sure there is enough maintenance in the future for this because the state in 0.8 was partially broken. Besides all this we currently think about switching the web frameworks (e.g. for 1.x) which could result in a complete Servlet-less way of development. E.g. using dropwizard or Vertx which would make thinks easier for contributors.
What is the reasoning behind the war deployment? Why would elastic beanstalk being easier?
E.g. instead of Tomcat there seems to be pure Java supported too. What would be better?