You can only set email preferences for owned, reported, QA or CC's bugs. What happens to voted bugs?

Marking as dependent on 71793 because with any luck both will get fixed together.

Currently voted bugs get all mail in the old format. Adding the pref framework for "When I voted on a bug" should fix that problem at the same time (one of the causes for bug 71625).

Marking for 2.16, but the votes getting old emailtech should definitely be fixed for 2.12. Is there a separate bug for that? I'm a bit behind on all the email filtering bugs.

*** Bug 62751 has been marked as a duplicate of this bug. ***

This bug can't be fixed until bug 73665 is because any new filters being added will default to "off" (this was discovered on one of the bugfixes to the original bug 17464).

In order to get this in for 2.12 (part of bug 71625), I've added a but of a kludge to the code. If a value is not defined or 'on' for a filtering group, it means mail should be delivered. If it is defined and is "", then no mail is delivered. Basically, this makes it so that new options default to 'on'. The patch also impliments the filter group "Voter".

Jake, your "kludge" to processmail seems to make the code tolerant against missing userFlags. In my view, this is the right thing to do, and you could drop the printed warning. Please compare this to attachment 27389 [details] [diff] [review] (my proposed fix for bug 71600) and attachment 27405 [details] [diff] [review] (the "simplified" version of that patch that got checked in): My approach was to check for "on" and "off" separately, and exclude the person from getting mail iff "off" was found and "on" was not found. Your approach is to treat missing values as "on" and exclude the person from getting mail iff "on" was not found. I think both solutions are correct and equivalent. Yours may even be cleaner. In any case, you have r=afranke on that "kludge" :) The rest looks good to me, but that doesn't say much...

I call this a kludge because the logic behind what it has to do is not pretty. For someone looking at the code for the first time, the idea of treating undefined the same as 'on' will seem quite strange. At least it does to me. The proper way to fix this (IMHO) is bug 73665, but that is a bit much for 2.12. And yes, I guess this does do the same thing as the part of attachment 27389 [details] [diff] [review] that got left out (although this one adds the same logic to userprefs.cgi so the checkmarks agree w/the filtering).

I agree, the warning isn't needed. This is the right thing to do, since the prefs should default to 'on' if they haven't been defined yet. This is the same behavior it originally had when they were converted to the new pref system, so it only makes sense to do that when adding prefs. r= justdave Checked in (without the warning line).

Moving closed bugs to Bugzilla product