Github user koeninger commented on the pull request:
https://github.com/apache/spark/pull/3798#issuecomment-72773257
To be clear, I'm ok with any solution that gives me access to what I need,
which in this case are offsets.
What's coming across as me feeling strongly about it is, I think, really
because I'm not clear on what your principles with regard to the interface
are ... well, that and frustration because it's been changed 3 times
already :)
For instance, why would you be willing to take on the fragile base class
maintenance problem in exposing KafkaRDD as something that could be
subclassed... but not ok with exposing the DStream (so that people could
override the batch generation policy)?
In the interests of moving this forward, if we're really just talking about
changing KafkaUtil's use of
RDD[..] with HasOffsetRanges
to
RDD[..]
I can make that that change
On Tue, Feb 3, 2015 at 7:02 PM, Tathagata Das <notificati...@github.com>
wrote:
> Okay here are the two options.
>
> 1. createRDD returns RDD[(K,V)] or RDD[R], and DStream.foreachRDD uses
> rdd.asInstanceOf[HasOffsetRanges]
> 2. createRDD returns KafkaRDD[(K,V)] or KafkaRDD[R] and
> DStream.foreachRDD uses rdd.asInstanceOf[KafkaRDD[_]]
>
> I think I am okay with either one. Stepping back, my original concern was
> returning something that had no binary compatibility issues. Both solution
> suffices. Between these two, since you feel so strongly against (2), lets
> go with (1).
>
> â
> Reply to this email directly or view it on GitHub
> <https://github.com/apache/spark/pull/3798#issuecomment-72770451>.
>
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
