Github user paul-rogers commented on a diff in the pull request:
https://github.com/apache/drill/pull/993#discussion_r144930316
--- Diff:
exec/java-exec/src/main/java/org/apache/drill/exec/server/rest/WebUserConnection.java
---
@@ -124,6 +124,13 @@ public void sendData(RpcOutcomeListener<Ack> listener,
QueryWritableBatch result
}
}
+ /**
+ * Returns a DefaultChannelPromise which doesn't have reference to any
actual channel but has an EventExecutor
+ * associated with it. In this case we use EventExecutor out of
BitServer EventLoopGroup. Since there is no actual
+ * connection established using this class, hence the close event will
never be fired by underlying layer and close
+ * future is set only when the WebSessionResources are closed.
--- End diff --
Can we take this comment up a level? This is a description of the
mechanics, which is fine. But, what is the intent?
Are we using a web session as a pseudo-connection (as described in earlier
comments)? Is this an adapter to make the session look like a connection?
---
