[
https://issues.apache.org/jira/browse/DRILL-325?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13847574#comment-13847574
]
Jacques Nadeau commented on DRILL-325:
--------------------------------------
Hi Henrik and welcome. Its great that your interested in supporting MADlib. I
can't see why this would be challenging. Since it is very dear to your
interests, I'd suggest that you start coming up with a phased design plan for
how to support the various capabilities. We already had support for UDFs and
UDAs. Because our operators are pluggable, we could also implement an
iterating operator if that makes sense. Since MADlib is C++, we'll need to
implement some frameworks as we don't currently have any C++ UDF or UDA
examples. We'll also need to think about what aggregate parallelization means
in the context of MADlib as our current approach is that all workspace
variables that will be merged in the second phase of distributed aggregation
must be internal types. I'd love to see your thoughts on where and how you
think MADlib should be integrated & supported. As a community driven project,
we need help from experts like you to get these types of features completed.
> Support for MADlib
> ------------------
>
> Key: DRILL-325
> URL: https://issues.apache.org/jira/browse/DRILL-325
> Project: Apache Drill
> Issue Type: New Feature
> Reporter: Michael Hausenblas
>
> It should be possible to use MADlib (http://doc.madlib.net/latest/) with
> Drill.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
