[racket-dev] Responding to old bug reports

On Fri, Dec 2, 2011 at 4:39 AM, Eli Barzilay <eli at barzilay.org> wrote:
> Yesterday, Sam Tobin-Hochstadt wrote:
>> In my role as bug czar, I've been trying to work through the backlog
>> of old unexamined bugs.  Unfortunately, many of them (a) have very
>> little information and (b) are from a long time ago.  A good example
>> is this one: http://bugs.racket-lang.org/query/?cmd=view&pr=9481
>>
>> There are a few options here:
>> 1. Close old (how old?) bugs that don't have enough information to
>>    triage/reproduce.
>> 2. Put them in the "feedback" state, with a general request for more
>>    information.
>> 3. Attempt to guess at what component this belongs to, and assign it
>>    that way.
>> 4. Leave it in the current state.
>>
>> I think 3 and 4 are both unacceptable. I'm leaning toward 2.  The
>> major drawback of 2 vs 1 is that we accumulate lots of bugs that
>> nothing will ever happen with. The major drawback of 1 is that we're
>> saying to someone who took the time to report a bug "we didn't care
>> about this bug at the time, so now we're closing it".
>
> I think that a sensible choice would be to combine #1 and #2: decide
> that PRs that have been in feedback state for a while are retired.
> It's easy to do this -- here's a query that shows all of these PRs
> sorted by date of last modification:

That's basically what Robby suggested as well, and it seems right to
me as well.  I also did just close the example bug I gave.

> http://bugs.racket-lang.org/query/?State=feedback;columns=Category;columns=Synopsis;columns=Last-Modified;sortby=Last-Modified;cmd=submit%20query
>
> Looks like there are very few of them and most could be closed.

Unfortunately, some of these need to be re-triaged. For example, 6285
is a real bug, and I think is only accidentally in the feedback state.
 But there aren't that many, so it's ok.
-- 
sam th
samth at ccs.neu.edu