Edward,
Type-safe API currently is mainly for better development experience, not
fully general-purpose monitoring.
For example, when the developer is required to implement groupBy "user",
the implementation have to override lots of storm-specified code like:
https://github.com/haoch/incubator-e
This is a nice progress. good to see we have type safe fluent API.
Could you please explain a little bit more how typesafe would help generic
purpose monitoring?
In general purpose monitoring, do we need input type/output type defined?
How user describe their types from source to each processing
JIRA: https://issues.apache.org/jira/browse/EAGLE-66
Pull Request: https://github.com/apache/incubator-eagle/pull/17
*Main Changes*
1. *Typesafe API:* Currently the stream processing API is not type-safe
(Type info are erased by stream framework), all programming interfaces for
developer are face
Missed another change about *Configuration*:
https://github.com/haoch/incubator-eagle/blob/EAGLE-66-TYPESAFE/eagle-core/eagle-data-process/eagle-stream-process-api/src/main/scala/org/apache/eagle/datastream/core/Configuration.scala
*val env = ExecutionEnvironments.get[StormExecutionEnvironment](a
*TODO*
✓ Typesafe API
✓ groupByKey
☐ Test compatibility for old application
☐ Refactor package to seperate core/storm/spark/others
☐ Support KeyValue based grouping for old application and existing executor
/ spout
☐ Support groupByKey(fields: String*)
☐ Refactor Alert Executor with groupByKey and