[
https://issues.apache.org/jira/browse/MAHOUT-1570?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14579260#comment-14579260
]
Dmitriy Lyubimov commented on MAHOUT-1570:
------------------------------------------
I would like to emphasize that this work gets us all in Mahout totally excited.
Yes, with the current state of the code I personally would encourage you to do
a PR.
I think [~ssc] is missing the point that PR is tool of collaboration but not
just "request to commit". Making suggestions on PR is no different than making
suggestion on Jira, except it is more powerful since one could "point and
elaborate" on the code. Nobody will rush to merge your work prematurely and
until you say so, but it enables collaboration on the issue to a greater extent
than just published github branch or Jira comments -- IMO.
Ultimately it is up to you to if you want to limit the scope of external input
and collaboration before you present things to public. This is a totally valid
approach if you want to stay 100% focused on your work.
However it seems since you have published some ongoing work on this Jira, you
are at least somewhat ready to interact with reviews and suggestions. If so, PR
would alleviate that to a greater extent than this Jira.
Another reason to do PR is if we eventually all agree that something needs to
be changed, it is less work to do it when things are still fluid rather than
when things are all nailed down.
> Adding support for Apache Flink as a backend for the Mahout DSL
> ---------------------------------------------------------------
>
> Key: MAHOUT-1570
> URL: https://issues.apache.org/jira/browse/MAHOUT-1570
> Project: Mahout
> Issue Type: Improvement
> Reporter: Till Rohrmann
> Assignee: Suneel Marthi
> Labels: DSL, flink, scala
>
> With the finalized abstraction of the Mahout DSL plans from the backend
> operations (MAHOUT-1529), it should be possible to integrate further backends
> for the Mahout DSL. Apache Flink would be a suitable candidate to act as a
> good execution backend.
> With respect to the implementation, the biggest difference between Spark and
> Flink at the moment is probably the incremental rollout of plans, which is
> triggered by Spark's actions and which is not supported by Flink yet.
> However, the Flink community is working on this issue. For the moment, it
> should be possible to circumvent this problem by writing intermediate results
> required by an action to HDFS and reading from there.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
