Re: Bits from the Debian Pure Blends Team

[Chris Walker <chrisw@chiark.greenend.org.uk>]

Holger Levsen <holger@layer-acht.org> writes:

> Hi Andreas,
> 
> On Samstag, 21. März 2009, Andreas Tille wrote:
> > > One suggestion though: currently you calculated the bugginess of a
> > > metapackage by the number of bugs their depnds have. I'd suggest to
> > > divide that by the number of packages that metapackage depends on :-)
> > to regard.  Please elaborate (perhaps in private / German?).
> 
> http://blends.alioth.debian.org/edu/bugs/common.html states that dependend 
> packages have "1 critical, 2 serious, 122 important, 528 normal, 175 minor, 
> 511 wishlist" bugs, but it does neither display or include into the metric 
> the number of dependent packages. 
> 
> And "1 critical, 2 serious, 122 important, 528 normal, 175 minor, 511 
> wishlist" bugs are a lot for 2 dependent packages, but not for 100 dependent 
> packages. So the metric shown is not really so useful.

It provides an answer to the question "Which metapackage needs most
help". Normalising by the number of packages would, I agree, give a
fairer assesment of the relative bugginess of the metapackages. 

I think there is merit in both scores. 

Chris

PS There is also scope for adjusting the relative "badness" of each
severity. I proposed an potential alternative scoring system last year
- but haven't had the time to try it out yet.