Approved. Calculate current metrics stuff first & update the LEP. -- jml, [2010-04-22]

Close Bugs, while Leaving Discussions Open

There should be a way to close bugs without closing down discussions in the bug report.

As a bug supervisor for a project
I want a way to reject bugs without shutting down continued conversation
so that bug lists only have relevant bugs for me but community members can continue discussing issues.

Rationale

Why do this now?

Bugs continue to be filed against Launchpad that experience status toggling wars as developers seek to remove bugs from Launchpad workflows but users want them re-opened for discussion.

What value does a new approach provide?

If we had a way to remove bugs from bug lists on Launchpad while leaving discussion open, both developers and users would be happy.

Stakeholders

Mark Shuttleworth
Any bug supervisor for a project

Constraints

Keep the number of statuses to a minimum
Developers must not be confused by the choices offered in the status menu
Marking a bug in this way must not have the social impact of closing the discussion
Marking a bug in this way must remove the bug from developer bug lists
It must be easy to find and review bugs marked in this way

Workflows

Current suggestions have been to add a status called OPINION. The new status would behave exactly as WONTFIX and INVALID behave now, including ACLs.

Success

We will know we have been successful when discussions continue independent of the state of the bug and without causing bug status to toggle needlessly.

Metrics

These are notes taken from a conversation between deryck and jml and later kfogel:

Flapping of status: What is average bug flap score right now? Versus flap score for bugs marked as OPINION? (A bug's "flap score" is defined as the number of times it flips from a closed to an open status. OPINION will be considered a "closed" status. We're also interested in whether flapping goes up or down in general after OPINION is introduced: assuming there's no other new feature in bugs that might affect flapping, we can assume that correlation is causation.)
30 days after OPINION feature goes live, find 10 bugs where discussion continued after OPINION status, and 10 where discussion continued after WONTFIX or INVALID, and look for patterns in quality of conversation
Percentage of bugs marked OPINION? (Versus percentage that formerly were marked INVALID or WONTFIX.)
How many unique people have used OPINION?

If after 3 months we have no evidence that this feature is having a good effect (statuses flap around less, discussions are better because people aren't offended), then we leave "OPINION" status changes in the bugactivity log, but change all current OPINION-state bugs to WONTFIX and remove the OPINION option from the status selector.

Notes: also pay attention to how many bugs are filed against the feature itself.

Stats are tracked at the following links:

Release Note

Bug Supervisors can make use of a new status, OPINION, which will mark a bug closed but is meant to indicate discussions are free to continue in the bug.

Thoughts?

There has been some email discussion between Deryck, Jono L., and Mark about the name of the status. OPINION seems the best anyone can come up with but is still problematic due to the fact that it's purpose is not that clear.

Status names that were discussed and rejected include:

POLICY
DESIGN
MOOT
DELIBERATE
INTENDED
STYLISTIC
FEATURE
OUTOFSCOPE
CONFAB
RAPSESSION
DIGRESSION

Status names that have been proposed and are still being considered include:

DISCUSSION
DISPUTED
ONGOING
INDISCUSSION
COMMENTING
CONVERSATION
DIALOG
DEBATE
TALK

IRC discussion

<cjwatson> noodles775: I'd like to work on fixing bug 294846, since it's annoying me.  Do you have a moment for a pre-impl chat?
<mup> Bug #294846: Setting to Won't Fix is ACLed but unsetting it isn't <ubuntu-qa> <Launchpad Bugs:Triaged> <https://launchpad.net/bugs/294846>
<wgrant> Is it actually desirable? It's unprecedented.
<cjwatson> How so?  We have existing ACLs
<wgrant> Nothing that prevents users from rectifying triage mistakes and reopening bugs.
<wgant> (there's of course the obvious problem that that sometimes isn't wanted, but...)
<cjwatson> The current pattern is: developer says they don't intend to fix a bug, and rather than trying to persuade them that they're wrong in comments, the user just slams it back to New
<cjwatson> repeat until very bored
<wgrant> Well, there's discussion about introducing a new status for that, which looks less obviously closed.
<wgrant> So users will hopefully not try to reopen it.
<cjwatson> That seems like a silly workaround to me
<wgrant> Unless Mark starts Won't Fixing my bugs without comment again :P
<cjwatson> users will sooner or later figure out that it means closed, and then we're back where we started
<wgrant> Hmm.
<cjwatson> reopening bugs rather than engaging in discussion with the developer is a community anti-pattern
<wgrant> Indeed.
<cjwatson> the Debian BTS even says in the mail you get when a bug is closed that if you think it's wrong you should discuss it with the developer, who'll reopen if needed
<cjwatson> that said I don't think people should be prevented from reopening Fix Released - Won't Fix is different because it's generally the result of a genuine difference of opinion, in which case simply reopening it is unlikely to be useful
<cjwatson> still, if it's controversial, then I suppose a pre-impl chat is premature
<wgrant> https://dev.launchpad.net/LEP/CloseBugsLeaveDiscussionsOpen
<cjwatson> aha
<cjwatson> bizarre summary though.  Discussion remains open on Won't Fix bugs too.
<cjwatson>  I'll just link to that from the bug, then
<wgrant> Yes, I thought that a little odd too.
<cjwatson> distinction between socially open and technically open, I suppose