Mailing-List: contact dev-help@flink.incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@flink.incubator.apache.org
Date: Tue, 12 Aug 2014 21:22:12 +0000 (UTC)
From: "Fabian Hueske (JIRA)" <jira@apache.org>
To: dev@flink.incubator.apache.org
Message-ID: <JIRA.12733601.1407878446722.67105.1407878532281@arcas>
In-Reply-To: <JIRA.12733601.1407878446722@arcas>
References: <JIRA.12733601.1407878446722@arcas>
Subject: [jira] [Created] (FLINK-1045) Remove Combinable Annotation
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit

Fabian Hueske created FLINK-1045:
------------------------------------

             Summary: Remove Combinable Annotation
                 Key: FLINK-1045
                 URL: https://issues.apache.org/jira/browse/FLINK-1045
             Project: Flink
          Issue Type: Improvement
          Components: Java API, Scala API
            Reporter: Fabian Hueske
            Priority: Blocker


Since the {{combine()}} method was extracted into an interface, we can check if a function is combinable by checking if the interface is implemented.

Hence, the {{Combinable}} annotation could be removed, IMHO.

However, this would change the API because {{RichGroupReduceFunction}} could no longer have a default implementation of {{combine()}} that can be enabled by attaching the annotation.

Right now, the {{Combinable}} annotation is just not necessary as the implementation of the interface tells the system everything it needs to know.
Removing the {{Combinable}} annotation would also make the handling of different combinable functions easier. For example a function that implements the interfaces {{GroupReduceFunction}} and {{FlatCombineFunction}} does not need to attach the annotation.

Since this is not a big, but API changing change, I would like to get it into 0.6 (if we decide to do it).

What do you guys think?


--
This message was sent by Atlassian JIRA
(v6.2#6252)