On Thu, Jan 12, 2006 at 03:11:58PM -0500, Joey Hess wrote: > Andrew Suffield wrote: > > Well it's nice in theory. The problem is that you have to set the > > threshold high enough to exempt glibc and dpkg, and when you do that, > > I have not yet found a metric that complains about any other packages > > (I've tried two or three times to invent one). > > I think the problem might be that the formula doesn't take the package's > installed base and/or age into account. The number of bugs in the BTS > tends to increase as both values increase without much connection to > the actual number of bugs in the package that affect many users, since > people eventually hit most of the edge cases, and those sort of bugs are > often the least likely to get fixed.
It might help, if there was only a good way to sample this information. popcon is pretty dubious, and I've got no idea offhand for a good way of detecting the age of a package (particularly when you consider package renames and changelog rotation and such). -- .''`. ** Debian GNU/Linux ** | Andrew Suffield : :' : http://www.debian.org/ | `. `' | `- -><- |
signature.asc
Description: Digital signature
