http://maven.apache.org/guides/introduction/introduction-to-dependency-mechanism.html
They can still use hadoop 0.20 with the 1.0 artifact because we are
producing the exact same jar with either dependency. If you look in the
chart on that page, if someone lists mrunit as a test dependency, our
provided hadoop dependency is not included. The user is then free to use
whatever hadoop version they want. If they use 0.23 they will a runtime
exception related to the interface/class Context object issue. If they
provided either hadoop 1.0 or 0.20 mrunit for 1.0 will work.
On 04/04/2012 10:34 PM, Brock Noland wrote:
I agree on all accounts. The person who helped me setup the profiles
setup three so I kept it at that. I figured there might be some people
still looking for an 0.20 artifact but 1.0 has been out long enough.
On Wed, Apr 4, 2012 at 10:16 PM, Jim Donofrio<[email protected]> wrote:
Why did we make a separate profile for hadoop 1.0.0, that jar is exactly the
same as the one for 0.20? Since hadoop is changing hadoop 0.23 to 2.0, maybe
it would be better to have only 2 jar's, one would be hadoop1 and the other
hadoop2?