|
|
+
Patrick Hunt 2011-11-30, 17:37
+
Camille Fournier 2011-11-30, 18:27
+
Patrick Hunt 2011-11-30, 18:42
-
Re: FYI: MR v2 issue with 3.4.0Mahadev Konar 2011-11-30, 19:20
Yeah,
We should fix that in mrv2. Since the test jars are publised now, its easy to fix that in mrv2. I plan to do that soon (will do that along with 3.4.1). mahadev On Wed, Nov 30, 2011 at 10:42 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote: > Someone is using an unpublished API (a private api to zk), we could > say "tough luck" or we could look at how difficult it would be to > provide this in a b/w compatible way. Likely we refactored this > Factory code and it's no longer available to the downstream in the > same way. Might be simple, might be hard, I was hoping Mahadev might > know more given his involvement with mr2. > > Patrick > > On Wed, Nov 30, 2011 at 10:27 AM, Camille Fournier <[EMAIL PROTECTED]> wrote: >> I'm puzzled, what are we supposed to fix here? Perhaps the correct fix is >> to figure out what their TestZkClient is doing and provide a contrib that >> works similarly and is kept up to date with versions. >> >> C >> >> On Wed, Nov 30, 2011 at 12:37 PM, Patrick Hunt <[EMAIL PROTECTED]> wrote: >> >>> It's an unpublished api but we still might want to address this in >>> 3.4.1/trunk to reduce downstream impact: >>> https://issues.apache.org/jira/browse/MAPREDUCE-3478 >>> >>> Patrick >>> +
Patrick Hunt 2011-11-30, 19:31
|