[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Low Averge Bug Counts

* Matt Zimmerman <mdz@debian.org> [011208 11:07]:
> On Fri, Dec 07, 2001 at 11:20:52AM -0800, Grant Bowman wrote:
> > Christian Kurz and I came up with some other factors that are measurable
> > and may need to be taken into account to increase the validity of the
> > data.  Just the average open unarchived bugs isn't very accurate though
> > I feel it's a good start.
> > 
> >   * severity (critical, grave, serious, important, normal, minor,
> >     wishlist, fixed) of bug reports 
> >   * status (open, forwarded, pending, fixed, done) of bug reports
> >   * age of bug
> >   * responsiveness of user (scan log for From: & Date: and compute)
> >   * responsiveness of maintainer (scan log and compute)
> > 
> > In project management there are formulas for prioritization and
> > scheduling that may be applicable.  I can pull out the information I
> > have.
> This is just the kind of reporting I had in mind when I prototyped a backend
> for debbugs which stored all of the messages and bug data in the database.
> The From and Date fields were to be extracted, and some attempt could be
> made to guess whether the message was sent by the maintainer, the submitter,
> or someone else.  However, there were questions about the reliability of
> PostgreSQL, concerns about overcomplexity, and other, non-technical reasons
> why it wasn't feasible.

I would be curious to hear more about the reasoning you refer to.  It
sounds like the benefit was getting pretty unclear given the level of
work required to make it actually work and still be maintainable.

> While I like the idea of giving hard-working developers a pat on the back,
> I'm not sure that it's a good idea to rate them so mechanically.  It's even
> less valid than judging real-world system performance based on artificial
> benchmarks.

Yes, I agree the present numbers are nowhere near perfect.  What I don't
want to do is "rate developers" exactly.  What I was intending was to
use the numbers we have as one indicator of activity.  In this context
these numbers do have validity even if they are not very good as a
general rating.

And I know that a pat on the back is worth infinitely more than no pat
on the back, even if the system for ranking needs some work.  I was
intending to acknowledge that in the write-up.

> I think that we might do better to consider a user- and
> developer-driven system, maybe more along the lines of the netfilter
> scoreboard:
> http://netfilter.samba.org/scoreboard.html

Wow, I hadn't seen this.  It looks very nice.  If anyone else has 
examples of similar systems, please post them.  I'll take a closer look.

At first glance, it seems like a fairly involved system to implement for
Debian.  This may be very good, but I was hoping to use whatever numbers
we have now for a "holiday pat on the back."


-- Grant Bowman                                   <grantbow@svpal.org>

Reply to: