Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/338#issuecomment-71595710
I think in the current state, this makes sense.
I wrote the interface like it was, because it would enable implementations
that does not compute/provide all splits on all machines. Think of a collection
input format where we send a different subset of the collection to all each
node (not supported in the runtime now, but might be at some point).
I guess we can realize something similar by splitting on the client and
then sending the sub set iterators directly.
---
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.
---
