Github user kayousterhout commented on the pull request:
https://github.com/apache/spark/pull/12060#issuecomment-213890432
In theory this seems like a good idea: there do seem to be lots of
unnecessary calls to submitWaitingStages(), and those calls seem somewhat
expensive since they repeatedly re-compute the stage DAG. I'm a little nervous
about the change though, because the DAGScheduler code is generally somewhat
brittle and doesn't have well-documented invariants / abstractions, so this
should be merged towards the beginning of a release cycle, ideally. I'm most
nervous about the change to getAncestorShuffleDependencies (which adds a lot of
complexity) so curious to hear why that is necessary.
---
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]
