Thanks all, the PATCH_BRANCH works for me. https://github.com/apache/hbase/commit/187ff190ed49e595200a3f60305b17b5947e4234
In HBase we always start from the master branch in the precommit jenkins job. 2016-10-25 0:26 GMT+08:00 Allen Wittenauer <[email protected]>: > > I've been out so catching up on email. > > There's a bunch of different options without actually interpreting > PATCH_BRANCH (which will likely fall apart if someone is doing a > development branch): > > * the hbase personality that hbase actually uses is already per branch. > so just edit it on those branches > > * have the test read input from mvn/pom.xml/control file in the branch > > * pull the test out of the personality and into user plugin dir for that > branch > > There's pros and cons to all of these methods. Some things to > keep in mind are: > > - ease of mass commits to multiple branches > - different build flags (e.g. 3.x will probably need > different settings) > - if users run this manually, how do they know what to do > (e.g., Apache Hadoop's yetus-wrapper) > > > > On Oct 20, 2016, at 7:19 PM, suraj acharya <[email protected]> wrote: > > > > Hi, > > I worked a bit using yetus and I saw that there is a variable > > PATCH_BRANCH which > > has the name of the branch you are currently running on. > > Now if you can use an if statement on that variable you will be able to > > change the version. > > > > Other users correct me if I am wrong. > > > > Hope that helps you. > > > > S > > > > -Suraj Acharya > > > > On Thu, Oct 20, 2016 at 8:41 PM, Duo Zhang <[email protected]> wrote: > > > >> The HBase community wants to drop the support of hadoop-2.4.x and > >> hadoop-2.5.x in HBase-2.0. This means we need to use different > hadoopcheck > >> versions for master branches when running precommit job. > >> > >> Now the hadoopcheck is invoke in the hadoopcheck_rebuild function > placed in > >> the personality shell script of HBase. > >> > >> https://github.com/apache/hbase/blob/master/dev-support/ > >> hbase-personality.sh > >> > >> If we can get the branch name in this function, it is possible to use > >> different hadoop versions for different branches. Will yetus send the > >> branch name as a parameter when invoking this method? Or is there any > env > >> variable that specifics the branch we want to test? Or is it possible to > >> run some git commands in this function(You know the problem is I do not > >> know the path of the local git repo..)? > >> > >> Thanks. > >> > >
