+1
My (potentially ignorant) understanding would be that the Spark
integration is a good candidate to separate out into its own release
cycle. I can't think of a reason we'd need to keep it in the main tree
in terms of HBase Java API (it would only rely on public API stuff, right?).
The lack of folks with time and expertise makes me think that a separate
release cycle makes me think that's the right call.
On 9/10/17 4:16 PM, Ted Yu wrote:
bq. revisit the possibility of an independent release cycle run out of an
additional ASF repo.
This seems to be more practical than the other alternatives.
On Sun, Sep 10, 2017 at 11:37 AM, Sean Busbey <[email protected]> wrote:
Hi Folks!
Our Stack recently gave an updated timeline on HBase 2.0 related
releases [1] that has us quickly approaching feature freeze for beta
releases.
Previously, we had a great discussion on what it takes for our
hbase-spark integration to be ready for release[2]. (see the summary
in the scope document I put up on [3]).
Unfortunately, since the start of August my day to day tasks have
pulled me away from focusing on chasing down the goals in the scope
document. Additionally it seems, to me at least, that other folks are
similarly focused on other things.
Is there anyone with enough cycles to work through the milestones
called out in the scope document in time for deadlines that AFAICT are
measured in a handful of weeks?
If not I suggest we either start pulling things out and aim for the
2.1 release, or revisit the possibility of an independent release
cycle run out of an additional ASF repo.
[1]: https://s.apache.org/wedV
[2]: https://s.apache.org/O53T
[3]: https://issues.apache.org/jira/browse/HBASE-18405
