Bug Gardening

Revision as of 14:00, 6 February 2017

Background

Bug Gardening (or Grooming or Triage) is the process by which projects:

1. confirm new bugs that have been submitted
2. gather more information about the bug when necessary
3. prioritize confirmed bugs
4. review bugs to ensure that they are being worked on
5. retire old bugs

This activity, however, is focused on retiring bugs that are no longer relevant (#5), primarily because that is more clear-cut than full-fledged bug triage.

Bug Triage Guides explain the rules for how to treat bugs within a project. Here are three examples of Bug Triage Guides:

• Gluster Bug Triage Guidelines and their accompanying Lifecycle of a Bug
• Bug Management - How to Triage (MediaWiki)
• Gnome Bug Triage Guide

Another good resource is this "bug etiquette" guide.

If the project you're working with has a guide for how to triage bugs, the students should read that first. If your project doesn't have a published bug triage guide, I suggest contacting the project to find out if there is an established mailing list or IRC channel for discussing bugs.

Directions

Though all bugs need to be triaged (see examples of Bug Triage Guides), this activity is focused on culling bugs that are no longer relevant, primarily because that is more clear-cut than full-fledged bug triage.

This activity is best performed in groups of two or three students.

Students should:

1. Read the bug triage guide for your project (if available).
2. Join the IRC channel or mailing list for your project on which bugs are discussed.
3. Identify a "suspect" bug in the bug tracker,
   1. Identifying the "suspect" bugs will be easier if you have information from the project about their bug lifecycle or which bugs they would prefer your students focus on. However, even in the absence of that information, there are some other clues that can be used to identify bugs that may no longer exist.
   2. Look at the version number of the software that the bug was reported against (versus the one that's current now). If the bug is for a version that's several releases out of date, there's a good likelihood that the issue may have been fixed (or may have just become irrelevant) in the meantime.
   3. Another possible hint is the date on the bug: bugs that were reported years ago may have been fixed and are probably worth looking at first.
4. Test the bug to see if it still exists. Once the student teams have identified a bug to work on, they should try to verify the bug status.
   1. At that point, if the bug is no longer relevant, your students can take one of several courses of action, depending on how well-established your/their relationship with the project is:
      1. If they have established a good working relationship with a project and have permissions to do so, they can close the bug (this is something a more advanced student might do)
      2. They can add a comment to the bug with their notes with the suggestion to close. This is more appropriate for students with less project experience.
   2. If, after testing, the student can not tell if a bug is still relevant -- that's ok. At that point they can:
      1. Move on and try another bug, or
      2. Ask for more information from the bug reporter or the person assigned to the bug. This can be a valuable because it exposes students to more of the project ecosystem.

Deliverables

At the end of the exercise, the student teams should have a list of bugs that they investigated and the outcomes. In the ideal case, they will have been able to close a bug or two, but even if they haven't you should still receive a report out. I suggest the following as a sample format (I've filled two lines as examples):

Assessment

Comments

If you're working with a specific open source project:

• ask the project (via IRC or mailing list) if there are a particular set of bugs that your students should focus on. For example, in the Gluster project I noted above, if a bug is tagged POST it means that an initial patch for a bug has been put into the Gerrit code review tool. If there is a very old bug that's marked POST, there's a decent likelihood that the patch has been incorporated into the project and the bug wasn't closed for whatever reason. If your students can verify this and close the bug, that's valuable.
• let the project know that your students will be doing this -- an influx of activity on a bug tracker can be overwhelming if unexpected

Other projects may have similar set-ups that are worth asking about.

Also: Make sure your project has enough old bugs for all your student teams :-)

Good Quiz Questions

1. Name some reasons that bugs that could have been closed might still be open?
2. Diagram the "normal" lifecycle of a bug for the project you're working on. Label the bug stages and who is involved at each stage.
3. Find several bugs from your project: one that should be closed according to the criteria you're using (that are in the project's Bug Triage Guide, for instance), one that needs more information, and one that has the in correct status. Have students triage them and evaluate their success based on the criteria they site in the assessment as well as the resolution they suggest.

Variation If you're working with a specific open source project and you have a block of time at least 1.5 hours long, you may want to run a bug sprint. You may be able to find a time when the regular "bug wrangler(s)" from the project can join you (via IRC or Google Hangout) and offer your students "live" assistance with triaging bugs.

Additional Information

Suggestions for the Open Source Project

Your community should have specific expectations around [support] that are published [somewhere]. For example, if your code will only work on Fedora versions newer than 19, then specify that.

If there are a set of bugs that it would be more helpful to have someone verify, then marking those in some way would help the instructor.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License