For example, I have a job and its results are below:
build5 - successful (change5 in SCM)
build4 - failed (chaneg4 in SCM)
build3 - failed (change3 in SCM)
build2 - failed when code was checkouting from subversion (change2 in SCM)
build1 - successful (change1 in SCM)
When I get changeset of build4 or build5 I can see change4 or change5. When I get changeset of build2 I see nothing and it's normal. But build2 didn't create revision.txt and If I get changeset of build3 I expect to see change3 + change2, but I see nothing too. Change2 and change3 dissappeared for jenkins completely.
I looked to source code of subversion plugin and saw method parseRevisionFile. This method has parameter findClosest.
All what I need is previousRevisions property assignment with findClosest=true instead of findClosest=false.
So I've replaced assignment from
previousRevisions = SubversionSCM.parseRevisionFile(build.getPreviousBuild());
previousRevisions = SubversionSCM.parseRevisionFile(build.getPreviousBuild(), true, false);
rebuilt plugin and it's work for me.
But I have questions: Is it a right way to lose any change information? Why is findClosest=false by default? Are there any problem with findClosest=true? Are there ways for use subversion plugin with findClosest=true without rebuilding plugin?
Thanks in advance.