[Libreoffice-qa] Stagnant NEEDINFO bugs

[Petr Mladek]

On Mon, 2013-01-28 at 10:05 +0100, Rainer Bielefeld wrote:
> Petr Mladek schrieb:
> > This will cause many mails only in the first round. It will be normal
> > level of mails if we do this regularly.
> 
> 
> Hi,
> 
> That's an illusion, total number of mails will always be the the same. 
> Only the number of mails per cleanup will be smaller.

IMHO, there is a difference when you get 100 mails now because we want
to clean up the current "mess" or when you get 5 mails per week when we
do this regularly.

> BTW, I dislike the "noise" the discussed "3 strikes" solution will 
> cause. I'm thinking about a different solution:

I am against 3 strike solution as well :-) My opinion is that it would
cause to big traffic and do not help much. If people does not react for
the first warning, there is only small chance that they would react on
the second or third one.

> Strike 1:
> Query will find NEEDINFO bugs untouched for a long time and fulfilling 
> some additional "hopeless criteria".
> Reporter's of these bugs will get polite mail with request to contribute 
> additional info that we will have to close the bug without additional 
> info. This mailing  only send mails to reporters, will not change any 
> info in the Bugs, so that data as "Days since last change" and similar 
> will be available for other queries. List of related bugs will be 
> published on QA list
> 
> That's not a big technical challenge, I think I can create required 
> tools (what can be used fur further actions in future easily) within 1 hour.
> 
> Strike 2 After 7 Days:
> Query for all Bugs for what mails have been sent in Strike 1:
> - Changed since mail (probably by reporter): QA will take care
> - NOT changed: Mass close via Bugzilla with polite message
>    "Sorry ..., but feel free to reopen if ..."
> 
> What do think?

I like this solution. It is polite and creates only one change in
bugzilla.

> BTW, I would not do that too often. Sometimes it's simply not easy for 
> reporter to contribute desired info, for example because bug is not 
> simple to reproduce. May be such bugs can be marked by entry of a QA 
> "Mentor" in QA contact or similar.

I would do this regularly to keep bugzilla clean and avoid masschanges
in hunderts of bugs. There are different reporters, so we will not touch
the same reporter in each round. IMHO, the most important is to give
user chance to answer before the first warning (30 days or so). 


Best Regards,
Petr