Questions can be expired too early

You can always assign a question to link to a bug to ensure the bug does not expire. The question owner can reopen the question if he chooses, but it is not obvious that replying to the email will reopen it.

Curtis, that's not the same bug. Expiration in this case is wrong in the first place since there is activity on the question.

why is this issue high? This behaviour has existed for 6 years? I think other project understand to assign the question if they do not want it to expire, or users know to reply to which Lp knows to reopen. This does not seem to be a high issue for projects that have used answers for many years. What team can commit to changing this behaviour to justify high?

Hi Julian.

Here is some information about QuestionSet.findExpiredQuestions(). The query discounts assigned questions and those linked to bugs. It looks for questions where the Question.datelastresponse and Question.datelastquery are both older than 15 days old.

In your example then, Lp believed that the neither you or the asker has pursued the question-answer dialogue since 2012-06-09. We can see the response on the 8th. So what was the response on the 22nd? I suspect the Dan used the Comment action, which I believe is an aside from the dialogue....https://answers.launchpad.net/maas/+question/199796/+history

The question's two crucial timestamps are set when the message is added via _newMessage(...update_question_dates=True). The default is true, The Question.addComment() action does not change the timestamps and it is the only callsite that passes False. We certainly do not want it to change the timestamps if the question is SOLVED. Maybe you want comments to update the stamps. Maybe you can remove the kwarg and make _newMessage and make it smarter...any message when the question is not solved updates the timestamps.

Thanks for the analysis Curtis. I marked it high because it's easy/trivial to fix and it's a very misleading piece of UI at best, and a really annoying misbehaviour at worst.