Josh Elser wrote:
Sean Busbey wrote:
On Fri, Dec 5, 2014 at 5:24 PM, Christopher<ctubb...@apache.org> wrote:
This vote fails, with:
-1: Mike, Sean, John
+1: Christopher, Keith
Even without my implicit +1, or Bill and Josh's late -1's, the vote
still
fails.
I think it would help if each person voting -1 (for reasons other
than the
fact that there is disagreement or the thread was confusing or
difficult to
follow), would please reply with a very short and concise summary of
what
guidelines they think we *should* adopt.
The compromise proposed by Josh that works for me is the original with
the
addition of
* No client RPC removals or change in behavior of extant calls prior
to 2.0
(non-inclusive or inclusive doesn't matter to me)
* Documenting the above in release notes for 1.x by way of tests that
clients compiled with 1.6 work when talking to a cluster running the
newer
release
Thanks *so* much for the clarity here. This greatly helps.
The only thing that is a concern about testing of 1.6 clients against
the newer release. Are you signing up to do said testing? If you fall
off a bridge/get sick/etc, I don't want all newer releases to be blocked
until another community member volunteers or you are free to work on it.
Is that reasonable?
Actually, I also (embarrassingly) don't know the compromise that I
proposed which you're referring to. If you can re-copy that for clarity,
that'd be wonderful.
