Re: Re: Stable code with unstable dependancies.

On 31 Aug 2003 at 1:39, Greg Beaver wrote: > Cipriano Groenendal wrote: > > Hello list. > [snip] > > Should something be done about packages like this? IMHO the > > installer should, with preferred-state set to stable only > > download > > stable packages, unless it's -forced. Even better would be if > > packages depending upon something unstable could never reach the > > Stable condition unless all its dependencies would be. The same > > requirements could be made for the Alpha/Beta/Devel levels of > > stability. > > I think the requirement for "stable" makes perfect sense - any stable > package must not have a required dependency on a less-than-stable > package. Optional dependencies are always of course allowed, as Alan > has fixed in DB_DataObject. The installer should in fact always honor > the preferred state, and it currently isn't in every case, something > the core devs are trying to track down. What to do if a package v1.4 is stable and v2.0 is again marked as "beta" because of many api-changes? With an optional dependency you would install 2.0 but with a required dep you would install v1.4? This leads us again to the multi-version-packagesystem *g* Stefan