What do we want to do to about continuous integration for 1.5 and trunk,
now that 1.5 has branched? There's a potential for a lot of noise from
duplicates due to merging bugfixes from 1.5 to trunk.
Some options, each with their own pros and cons are:
Build both, but batch merges.
Check for svn updates less often.
Build both, merge when a bugfix is complete, but batch builds for trunk
(like max: nightly).
Build only 1.5, until 1.5 is released.
Build both, but only have Jenkins report back to JIRA for one of them.
There's also the decision of whether to have a separate build for Hadoop 2,
for trunk and 1.5 also, so 4 builds total...
Any opinions? Personally, I'm fine with Jenkins being spammy.
Christopher L Tubbs II
Eric Newton 2013-02-20, 14:49
David Medinets 2013-02-20, 19:57