UsingLowerPrioritiesAndVoting
Better Prioritization of Issues
We have many open issues in NetBeans. Besides fixing all issues (which is a desirable but unreachable goal) one way to improve the quality is to focus on the more important issues first. For this we currently use priority - mostly we track and watch only P1-P3. This works fine for P1s and P2s - these are not so many and are generally kept under control. But the huge number of P3 means we have little information about which one of these are important and should be fixed first so we pick randomly. As a workaround for this in stabilization phase QA marks some P3 issues with "fix-in-55" in status whiteboard. Here are two simple proposals how to improve prioritization of P3 issues.
1) The first proposal is to use voting. There are issues with low priority according to guidelines which are annoying for many users. There are also issues with many duplicates which have the same priority as issues detected only by one user. If we start paying more attention our users will start casting more votes which will give us more feedback. Voting in IZ is not ideal now. You have 5 votes per module which does not make much sense - it would be better to limit the number for all of NetBeans but this is not easily doable. Also, moving an issue to another component deletes all the votes. Even with these problems voting is usable and can improve the quality.
2) Another proposal is to use priorities more than we do now - to use not only P1 - P3 but categorize the large number of P3s into P3 - P5. We can use P3s for more important issues that we would like to fix in the current release and P4 for other valid defects which are not as important. P4 and P5 are almost never used to report new issues and somebody might say that marking an issue as a P4 means that it will never be fixed. But the point of prioritization is that not all issues can be fixed and we should focus on the more important ones first. Of course this means that when we currently watch the number of P3s and use it as a release criteria we need to extend this and also use the number of P4s.
Not sure about the value of this proposal. I guess that developers already do these decisions before they go and fix a bug.
3) Another proposal for prioritization would be using $ instead of votes see: http://www.nbbounty.org/
