I am not fine with dropping it. I still run it in several places. Believe it or now many people still run 0.20.2. I believe (correct me If I am wrong) facebook is still running a heavily patch 0.20.2.
Mainly most of the things hive uses were designed around the 0.20.2 mapred api. In fact hive's components are generally layered on top of the mapred.* and nothing really uses mapreduce. I think hive should go on supporting 0.20.2 indefinitely. 0.20.2 is solid. If you look at the shim layer, it turns out that supporting 0.20.2 is really not much more work then supporting any other version. Since the burden is not that high I do not even see why we would need to drop it, there is just as much or more shim-ing to try to deal with differences between 2.205, 0.23, 0.2, 2.0. I could see dropping 0.20.2 if it was a huge burden but I do not see it that way, it work's it is reliable, and it is a known quantity. On Wed, Sep 18, 2013 at 4:22 PM, Ashutosh Chauhan <hashut...@apache.org>wrote: > I am fine with dropping support for 0.20 line. 4 years is a long time. We > cannot keep accumulating tech debt forever. > > Ashutosh > > > On Wed, Sep 18, 2013 at 1:04 PM, Brock Noland <br...@cloudera.com> wrote: > > > Hi, > > > > At present we require compatibility with Hadoop 0.20.2. See: > > https://issues.apache.org/jira/browse/HIVE-5313 > > > > Considering 0.20.2 was released 4 years ago, how long are we going to > > continue to support it? > > > > Brock > > >
