On 2016-10-19 10:50:17 +0200, Rene Engelhard wrote: > On Wed, Oct 19, 2016 at 10:20:38AM +0200, Vincent Lefevre wrote: > > I don't remember exactly, but unsatisfied Recommends as an effect > > on the resolver cost in aptitude: > > > > https://www.debian.org/doc/manuals/aptitude/ch02s05s05.en.html > > Erm, wait. You file a bug claiming it breaks upgrades based on > resolver costs (which do not matter for the actusl upgrade), so > purely on a theoretical basis?
I sometimes have blocked upgrades (packages marked as broken) with a message saying that a "Recommends" is not satisfied (while there is no such thing with "apt"), but I'm not sure what is the exact cause. > > Moreover, even if the user has the choice, he may prefer not to > > upgrade when a Recommends would be broken, because it could make > > Well, then he doesn't. (Then we would need to stay on oldstable, but..) No, in unstable, I generally hold packages for a while if a Recommends would be broken, because I know that rebuilds may take some time (e.g. during a transition). So, a buggy "Recommends" is annoying. -- Vincent Lefèvre <vinc...@vinc17.net> - Web: <https://www.vinc17.net/> 100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/> Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)
