Github user rawkintrevo commented on the pull request:
https://github.com/apache/incubator-zeppelin/pull/928
@bzz Thanks for the comments, It builds fine on my machine, I wanted to see
how it did against CI. Been out of town for a wedding over the holiday weekend
(in US Monday was a holiday). I didn't think anyone would look at it before
today, and apologies for not outlining things more clearly. Looking at the
fail logs, it seems like flakey fails, but I'll investigate further this
evening.
@Leemoonsoo also thanks for the comments.
I do have some documentation on this as well at:
https://trevorgrant.org/2016/05/19/visualizing-apache-mahout-in-r-via-apache-zeppelin-incubating/
I agree there are those two approaches, and will present a 3rd where
'Mahout' lives as it's own interpreter.
so approach (1) a patch, approach (2) documentation only, approach (3) a
stand alone interpreter.
In general, options 1 and 3 make integrating Apache Mahout into notebooks
more 'turn-key' and since one of the primary use cases of Apache Zeppelin is
data discovery analytics, this seems like a natural fit that would benefit both
projects AND end users. Cons of 1 and 3 are that they add code that must be
maintained; as far as invasiveness of patch and code which must be maintained,
1 is vastly superior to 3 (for 3 we would be basically copy and pasting the
spark interpretter over with this patch applied). I do agree with Moon's
comments above for paramter setting based on 'spark.mahout=true/false" this is
to bzz's point WIP.
So pros/cons of approach 2 is in essence the inverse of 1 and 3. The pro:
no additional code to maintain (though a blog post which must be maintained as
I've learned the hard way re my post on Installing Zeppelin with Spark and
Flink), the cons of which: while Zeppelin does make all of the steps required
fairly easy, the steps required to add Mahout support would be at least fairly
considered an "intermediate" Zeppelin skill level. This is a barrier to entry
to several Apache Mahout users who are excited to see this integration as a
means for visualization in Mahout.
My .02.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
