On Mon, Aug 24, 2009 at 3:10 PM, Greg Stark<gsstark@mit.edu> wrote:
> On Mon, Aug 24, 2009 at 8:03 PM, Alvaro
> Herrera<alvherre@commandprompt.com> wrote:
>> Personally I still think debbugs would suit us perfectly, but 1. I don't
>> have time to handle it, 2. nobody else believes this, 3. the debbugs
>> developers are not very interested in helping us use it.
>
> I've been shouting about debbugs forever too.
>
> It's completely email based so we could just treat it as a mailing
> list without having to go visit a web interface to stay up to date. We
> could add CVS/whatever hooks so whenever a commit message says it
> closes a bug it gets closed automatically. =A0Effectively it would
> require no operational changes for developers who would just
> participate on a mailing list and commit changes like usual.
>
> It also has a web interface so you can go see the history to see the
> pending bugs to work on and their history of course. But that's not
> how we're accustomed to working and I fear anything like bugzilla
> would require dedicated bug-wranglers like Bruce to keep the
> connection between the email discussion and the bug tracker going
> because the developers would ignore the web site and the bug reporters
> would be unaware of any mailing list discussion.
Well, I think we're dropping a lot of the bugs early in the process,
before they even get a response. Of course, a lot of the reason for
that is because many of the "bugs" are actually usage questions, user
error, completely lacking in relevant detail, problems with products
other than Postgres, and/or feature requests dressed up as bugs to
make us feel guilty about them. It seems to me based on my short
tenure reading this mailing list that when someone provides a
reproducible test case of Postgres verifiably DTWT it usually attracts
plenty of attention and gets dealt with relatively quickly, usually
with a friendly "thanks for the report!".
There's nothing technological that prevents someone from quickly
finding a list of the bugs that haven't gotten a response at all, as
evidenced by the fact that I just did it upthread in less than 5
minutes going back to bug ~4900. The details of how to reproduce this
are left as an exercise to the student (hint: threaded mail reader).
Finding the bugs that were discussed but not resolved is harder, but
probably a lot less worthwhile. A lot of those are exchanges of the
form:
Report: your product has a huge problem that would affect nearly every user
Response: uh, we doubt it, because we'd've noticed that. can you
provide a test case?
<still waiting>
I think there's definitely room for some better bug wrangling, but
given the number of garbage bugs, the effort/reward ratio is likely to
be pretty high.
...Robert
