[
https://issues.apache.org/jira/browse/PHOENIX-848?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13976292#comment-13976292
]
Jeffrey Zhong commented on PHOENIX-848:
---------------------------------------
{quote}
Would it be possible to do the following (albeit more hacky fix) instead of
backporting HBase-9754
{quote}
It's quite doable. The issue of this approach will come when we have new code
which relies on the setReadPoint for 0.96. Because 0.98&trunk doesn't require
to call setReadPoint, likely after a while we will forget this and introduce a
hard to detect bug for 0.96.
{quote}
it's unlikely that there will be another 0.96 release
{quote}
[[email protected]] could comment on this one. I think we'll have new 0.96
release every month.
> Enable HBase 0.96 on Master & 4.0 branch
> ----------------------------------------
>
> Key: PHOENIX-848
> URL: https://issues.apache.org/jira/browse/PHOENIX-848
> Project: Phoenix
> Issue Type: Improvement
> Affects Versions: 4.0.0
> Reporter: Jeffrey Zhong
> Attachments: EnableOnHBase0.96.patch
>
>
> Currently Master & 4.0 branch can support HBase0.98 & Trunk. While there are
> basically two areas which make the code can't work on hbase0.96 such as HLog
> & region row lock change(MultiVersionConsistencyControl.setThreadReadPoint is
> removed from hbase0.98+)
> It seems better that we could create a layer to abstract out those hbase
> differences(as the differences aren't much) so Phoenix can also run on
> HBase0.96.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
