Github user pwendell commented on the pull request:
https://github.com/apache/spark/pull/2350#issuecomment-55308825
Hey all - regarding backwards compatibility. I agree we definitely need to
preserve all of the publicly documented interfaces, including environment
variables etc. And I think that even includes the original yarn command line
`Client` which was both undocumented and deprecated in Spark 1.0 and is
subsumed by spark submit.
To me though, preserving programmatic linking against the `Client` class
doesn't make sense - it was never intended or documented to be used this way.
Really it's an oversight that this was even public in the first place, but I'm
guessing it was just because it was assumed to be self-evident that this exists
for parsing command line options, etc and not for linking against in a program.
The classes are not in our public java doc, etc. Finally, AFAIK we've already
changed a bunch of things in e.g. `ClientArguments` in both Spark 1.0 and
1.1... including renaming that could break access to it from earlier versions
of Spark. So IMO this already isn't a public API and belongs as
`private[spark]` (even before Spark 1.0). I am indeed curious if people are
actually trying to use this.
---
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
