Roadmap updated with feedback

roadmap.md

@@ -1,22 +1,50 @@
-Eventing Roadmap
+# Eventing Roadmap
 
-# 0.1
+At the end of each milestone, there should be a new or updated demo linked from
+the [README](README.md).
 
-A. Ability to register an event sources and specify which events may be generated from that source [Proposal](https://github.com/elafros/eventing/issues/39)
-B. kubctl command to set up a trigger where a specific event will cause a specific action
-C. Example source generating an event
-D. Example action
-E. Events conform to CloudEvent 0.1 specification
+## 0.1 CloudEvents Demo
 
-# 0.2
+1. Ability to register an event sources and specify which events may be
+   generated from that source
+   [Proposal](https://github.com/elafros/eventing/issues/39)
+1. kubectl command to set up a trigger where a specific event will cause a
+  specific action
+1. Example source generating an event
+1. Example action
+1. Events conform to [CloudEvents](https://github.com/cloudevents/spec)
+   0.1 specification
 
-A. A local event can trigger a Google Cloud Function
-B. A Google source can trigger a local action
+## 0.2 Friendly Repo
 
-# 0.3
+1. Easy installation [issue#51](https://github.com/elafros/eventing/issues/51)
+1. We have clear guidelines for how we use Prow
+1. All the contributing info is tested and someone new to the repo has made a
+   PR by following the directions
 
-A. Pipeline the output of one action into the input of another
+## 0.3 Functions
 
-# 0.4
+1. An event can trigger an Elafros function [issue#52](https://github.com/elafros/eventing/issues/52)
+1. A local event can trigger a Google Cloud Function
+1. A Google source can trigger an action local within the Elafros cluster.
+   Likely this will use some scaffolding on the Google side to validate the
+   interaction between systems, such as deploying a Cloud Function that
+   forwards the Google Event to the Elafros cluster.
 
-A. an action may accept a stream interface to process many events at once
+## 0.4 Event Persistence
+
+1. Events can be stored in a buffer / queue
+1. Failed event handling can be retried
+1. An Event Source can be added to a cluster without recompile
+1. There is a well-defined interface that allows the use of popular event
+   brokers for persistence, such as Kafka and Rabbit Mq
+
+## 0.5 Pipelines
+
+1. Pipeline the output of one action into the input of another
+1. Pipelines can be a series of Functions
+
+## 0.6 Streams
+
+1. an action may accept a stream interface to process many events at once
+1. Events are delivered in order