Github user squito commented on the pull request:
https://github.com/apache/spark/pull/5400#issuecomment-108090271
> Hey, one idea for avoiding tests that write huge files: can we use a data
generator + streaming
> checksum function to be able to detect data corruption without having to
materialize the entire file?
I don't see an easy way to do that. `WrappedLargeByteBuffer` doesn't have
any hooks to generate the underlying byte buffers on the fly. This was
intentional, to keep the implementation as simple as possible in the initial
version, and to leave that as a potential optimization in the future. I could
add another implementation of `LargeByteBuffer` just for testing, but that
seems to defeat the purpose of testing the implementation we'd actually use in.
Or we could change the actual implementation to support that, but again, I'm
reluctant to over-complicate the initial version -- we could certainly do
fancier things but its not needed to just eliminate the hard barrier at 2GB we
have now.
I'm sort of just playing devil's advocate -- I totally understand the
hesitation on including the tests as is, but not sure what else we should do.
Definitely open to continue discussing it & other ideas
---
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]
