On 11/12/13, 1:26 PM, Sean Busbey wrote:
On Tue, Nov 12, 2013 at 2:48 PM, Josh Elser <josh.el...@gmail.com> wrote:
What about the other half: "encouraging" users to lag (soon to be) two
major releases behind?
I don't think our current user base needs to be encouraged strongly to
upgrade. And as I said previously I think this change provides them with an
upgrade path that's easier to stomach, but I suspect this is a point we
disagree on.
Yes, I believe that disagreement to be accurate :). IMO, HDFS WALs are
themselves a sufficient feature to strongly encourage users to upgrade
from 1.4 to 1.5 when running Accumulo in a production environment.
However, I do want to say that my feelings are not solely influenced by
1.5 features that 1.4 does not have. Providing avenues for users to use
old versions of code is extremely degrading to the morale of developers
who are trying to innovate and make things better upstream.
Having to repeatedly answer questions about why things are the way they
are or why they are sub-optimal with "If you upgraded, it wouldn't be an
issue" can be very depressing to a growing community. There are multiple
amazing devs who have invested crazy units of effort into making 1.5 and
(soon) 1.6 be a reality -- I want to make sure as a community we
encourage our users to reap the benefits of that work, and let
aforementioned devs see successes from their effort.
Also, users still aren't precluded from using 1.4 if they so choose (and
if critical bugs are found, it's very likely for them to be patched in
1.4). I don't see it as unreasonable to force users to upgrade to a
newer version of Accumulo to use a newer version of Hadoop.
