On Tue, Nov 29, 2011 at 10:20 PM, Stack <[email protected]> wrote:
> The first hbase 0.92.0 release candidate is available for download:
>
> http://people.apache.org/~stack/hbase-0.92.0-candidate-0/
Took RC0 for a spin in Bigtop. Here are the findings so far:
1. tarball has an extra level of subdirectories. this is subtle,
because that level happens to be a ./ but it can trip up tar
in certain situations so perhaps its better to fix it (I suspect
changes are required at the level of assembly spec).
2. This is an extreme nitpick, but it would be nice to have
executable bit set ONLY on real executables (things in bin/)
Currently it is set on docs and quite a few other things.
3. This is a little bit more sinister and it is not clear what
the resolution path here is. When compiled against .23
the MiniMRCluster dependency latches onto artifact
hadoop-mapreduce/test. There are 2 problems with that:
1. this works only accidentally (but as long as it does
may be it is fine) since we don't have an explicit
dependency on hadoop-mapreduce test artifact (nor
should we, I think!).
2. MiniMRCluster from there is soon to be deprecated
(MAPREDUCE-3169) ans HBaseTestingUtility should
really transition onto using MiniMRClientClusterFactory
to get a MiniMRCluster.
> I've posted a secure and an insecure tarball. Pick your poison.
Are there instructions on how to play with secure tarball posted
somewhere? Deployment, etc?
Thanks,
Roman.
