Hi,
If you wonder why occasionally we have broken Jenkins build, I can give you
one reason.

Look at the following import in TestMasterFailover:
import org.apache.hadoop.hbase.*;

When a new test was added which used ClusterStatus, there was no import to
be added in TRUNK because of the above import.
However, in 0.92, there is no such wild card import - so ...

I think we should refrain from using wild card import.

Cheers

Which check-in rules have we set up for our SCM? -Thomas

On 2/22/12 12:46 PM, "Ted Yu" <[EMAIL PROTECTED]> wrote:

>Hi,
>If you wonder why occasionally we have broken Jenkins build, I can give
>you
>one reason.
>
>Look at the following import in TestMasterFailover:
>import org.apache.hadoop.hbase.*;
>
>When a new test was added which used ClusterStatus, there was no import to
>be added in TRUNK because of the above import.
>However, in 0.92, there is no such wild card import - so ...
>
>I think we should refrain from using wild card import.
>
>Cheers

+1 on avoiding using wildcard imports. Eclipse's "Organize Imports"  
command gets rid of them pretty well.

Thanks,
--Mikhail
On Wed, 22 Feb 2012 12:46:47 -0800, Ted Yu <[EMAIL PROTECTED]> wrote:

> Hi,
> If you wonder why occasionally we have broken Jenkins build, I can give  
> you
> one reason.
>
> Look at the following import in TestMasterFailover:
> import org.apache.hadoop.hbase.*;
>
> When a new test was added which used ClusterStatus, there was no import  
> to
> be added in TRUNK because of the above import.
> However, in 0.92, there is no such wild card import - so ...
>
> I think we should refrain from using wild card import.
>
> Cheers