Re: Fundamental flaw in bug reporting system
On Thu, Jul 13, 2006 at 08:44:57PM +0100, Ian Jackson wrote:
> I don't think picking a package and comparing bug reports like for
> like across two distributions is `anecdotal evidence'. Anecdotal
> evidence is statements like `well I tried to submit a bug report and
> was discouraged'.
Yes, I stand corrected. The data you've given is not anecdotal, it's just
too narrow to base any particular conclusions on. (IMO)
> What is your opinion based on ?
My opinion is that your opinion isn't thoroughly researched. This is
partially due to the fact that you are asking the wrong questions.
To repeat myself, the questions to ask are, how quickly are (real) bugs being
reported, and how quickly are they getting fixed.
I agree that intuitively, it seems that if you have more bad bug reports,
then developers will waste more time dealing with them. But you need to
balance that with the possibility that making it easier to submit bug reports
may also cause problems to be reported more quickly. It doesn't really
matter how much free time a developer has if he doesn't hear about a problem
until 2 weeks after it's started affecting our users.
> > The amount of noise in the system is really a secondary concern if it leads
> > to bugs getting reported faster and fixed faster.
> And it is a primary concern if it leads to bugs being introduced more
> often or fixed more slowly because the developers are too busy dealing
> with bad bug reports.
I agree. So why not find out if that is really happening, before drawing
conclusions? There should be a way to do this empirically.