Github user JoshRosen commented on the pull request:
https://github.com/apache/spark/pull/8979#issuecomment-149660588
Hey @ehnalis, I like the core idea behind this patch. However, I'm a bit
concerned about how many files it needs to touch and the level of complexity
here.
I think that your approach 1, of creating a new ShuffledRDD variant to do
value-only shuffle, actually won't be too much work. I did something similar,
but optimized only for Spark SQL's `UnsafeRow` format, in #7456. Could you take
a look at the implementation of `ShuffledRowRDD` in that patch to assess
whether you could do something similar (but more generic) here?
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
