> On 09/11/2010 03:04 PM, Ciaran McCreesh wrote: > > Or does the problem only occur if you mix keywords and ignore > > dependencies? > > I think that if a package doesn't work in a mixed environment, that > points to a likely dependency problem. Sooner or later there is a good > chance it will bite somebody. > > Personally, I try to keep package dependencies correct. If a package in > unstable needs a library version in unstable, I depend on that version - > not on the library itself. Then we won't get burned in six months when > I forget all about this or am not around and things start going stable > in the wrong order. > > Sure, if the issue is something really exotic maybe we should just say > "don't do that," but usually there is a better fix. > > Personally I welcome these kinds of bugs, as they're the easiest way to > uncover non-obvious dependency issues that might otherwise make their > way into stable. Maybe we can't fix them all, but we ought not to just > dismiss them out of hand. I certainly wouldn't want to see the > bug-wranglers screening for them, for instance. > > Rich
++ There should be nothing stopping a user from running a mixed arch/~arch system. Those problems just point to our dependency information not being recorded correctly. It might be understandable that this info can be incredibly hard to get correct but that doesn't mean it isn't a valid bug. - Alistair
