fine-tune delivery of duplicate notification mails

Thanks for your bug report.

I'm not yet sure how best to address the underlying issues here, but your proposal sounds like a potential solution.

Still thinking this one through, but there are some other issues to consider (this is basically a "Dear LazyWeb" comment):

  * It's insufficient to consider only directly subscribed users as the people who are truly interested in a bug, because bug contacts and assignees are indirect subscribers too

  * In our current mail system, duplicate notifications are batched with any other changes made around the same time, by the same person

  * Even if they weren't clumped together with other changes, our model currently has no way to distinguish one kind of change notification from another

umm, im not subscribed to ANY bugs right now (so says launchpad "subscribed" section in my profile) yet I get TONS of mail about dupes to some old bug I submitted :/

I have added an X-Launchpad-Message-Rationale header via bug 3797; this will let you filter mail based on whether it comes in via a duplicate. It doesn't solve this problem may it may mitigate it enough for it to go away by itself. Unfortunately it will only affect beta right now.. soon to be known as 1.0 though :-)

If a bug attracts over 15 duplicates is it possible to add a note to the bottom of the email indicating steps of how you can unsubscribe from it?

As mentioned in Bug #121434 the mass removal of the apport crash dumps can cause mailbox crushing too - perhaps mail about this should be ignored?

I don't think it's very useful for a reporter of a bug to receive notice of duplicate bugs. Surely this is only of interest to whoever is trying to correct the bug?

To stop all the extraneous emails
1 Unsubscribe from the bug you originally reported.
2 In Launchpad click on "Advanced Search" scroll right to the end & unselect "Hide duplicate bugs", click search and unsubscribe from all duplicates.
3 As a crosscheck click on the launchpad link in a duplicate email, if their is no unsubscribe option then there should be no more emails. If there is an unsuscribe option - do so and go through the rest of them unsubscribing.

Hope this helps

I think VJ is on the right track; specifically, I don't think subscribers to a bug should get email notifications if other bugs are marked as duplicates of that one. I think in some cases it is marginally useful; i.e. you want to look for more information in the duplicates, or you want to make sure that the duplicate is in fact a dupe -- but in the pathological cases it just annoys the hell out of everybody.

Does anyone want to make a strong case for why we shouldn't effect this change?

Whether knowing that a bug was marked as a duplicate of a bug you're subscribed to is interesting is determined by the amount of information in the duplicate, and to some extent its age.

If a an old bug with many comments is marked as a duplicate then the subscribers of the target bug might want to know about it so that they can have a look and see if there's any helpful information. For bugs that are filed and soon after marked as duplicates, this is indeed annoying.

I'm not sure I'm against canceling those emails altogether, but if we wanted to preserve some of their utility we could conditionally only send an email if the duplicate has comments, and is older than one day.

Knowing about a new duplicate being marked *might* be of interest to a developer or triager, to let them know that there's activity on the bug even if there aren't new comments.

It is *almost never* interesting to anyone else subscribed to the bug (including reporters of other duplicate bugs).

I would support suppressing these emails entirely.

I'd support suppressing them as well.

There is some marginal utility to know about the activity, but it's pretty tiny, and in pathological cases can create a pretty hefty amount of unwanted email. I'd say after about 2 or 3 such notices, whatever marginal utility is gone, and after about a dozen it's pure annoyance.

This will become especially important as we a) encourage people to file bugs separately more often, and b) employ automated tools like apport to capture and auto-dupe bugs.

2009/6/16 Matt Zimmerman <email address hidden>:
> Knowing about a new duplicate being marked *might* be of interest to a
> developer or triager, to let them know that there's activity on the bug
> even if there aren't new comments.
>
> It is *almost never* interesting to anyone else subscribed to the bug
> (including reporters of other duplicate bugs).
>
> I would support suppressing these emails entirely.

Me too.

--
Martin <http://launchpad.net/~mbp/>

+1 for not notifying duplication events by default. This happens frequently in GNOME bugzilla and there is always angry email returned. I would be very skeptical if any of the bug subscribers are even interested in the fact that duplicates are being reported anyway - if the duplicates contain any useful information then this is normally added to the parent bug in a comment and everyone is notified of this.

There is some value in developers/maintainers being notified of duplicates and this class of user can be expected to have a mail sorting scheme to handle the traffic. I'd recommend an opt-in option for duplication events.

In my experience I like being notified of duplicates on things I am maintaining and duplicate notices I get for bugs I filed 2 years ago I don't want to be notified about.

I find most e-mails Launchpad sends to be of little direct use to me, so I'd rather not get them. Being notified of duplicate bugs is not useful to me.

I'm also heavily in favor of suppressing emails like "bug 123 has been marked as a dupe" to non-direct subscribers. I get a lot of complaints from users for heavy crash bugs.

LP has become way too chatty with sending bug mails, such as those, or deleting an attachment for cleaning up, etc.

I'm convinced. Let's see if we can get to this in the current milestone, if not it should be doable in 2.2.7.

Fixed in devel r8937.