I see two points. 1) Only adding features to the hadoop-2 trunk. I guess most of us would agree on that part. +1 for me
2) Fixing bugs on the hadoop-1 trunk. The question I would have here is about versioning. We can't afford to keep the same version for both but if there is two different current versions then it should be explicit/obvious for the end user. Dave's proposal seems logical. hadoop1 -> 1.1.z : the version 1.1 is fixed and it is only minor bug fixes so only the z change hadoop2 -> x.y.z : the real active trunk of the project with bug fixes (z), new features (y) and major changes (x) I am afraid that I know hadoop1 clusters that are not likely to be upgraded soon. It seems a bit harsh to state that the hadoop1 version is not maintained anymore. Bertrand On Sat, Jun 7, 2014 at 11:32 AM, Dave Beech <[email protected]> wrote: > +1 > > Would we abandon development for hadoop1 completely or still produce > maintenance releases for critical bugs? No reason to keep the version > numbers aligned after all. We may have a 1.1.1 release for hadoop1, but not > a 1.2+ > > Dave > > > On 6 June 2014 21:56, Jarek Jarcec Cecho <[email protected]> wrote: > > > +1 > > > > Jarcec > > > > On Fri, Jun 06, 2014 at 01:51:21PM -0700, Brock Noland wrote: > > > Hi, > > > > > > Currently we maintain both trunk and trunk-hadoop1 so that we can > support > > > both Hadoop1 and Hadoop2. Given that users are moving towards Hadoop2 > and > > > we have just released, I believe we will have diminishing returns by > > > maintaining both branches. > > > > > > I propose we stop maintaining trunk-hadoop1 and move forward on trunk > > which > > > supports Hadoop2. > > > > > > Brock > > >
