On Tue, 2008-08-26 at 14:45 +0100, Simon Marlow wrote: > > We can hack it for the special case of base, or try and be slightly more > > general and not complain when the two versions that get pulled in > > directly depend on each other. In general there's no simple way for > > Cabal to know if using multiple versions of a package is going to work > > on not. It's quite unsatisfactory. > > If there's a dependency, direct or indirect, between the two versions of > the package then it is ok to include them in the same program. The only > way this could occur is if the two versions are intended to be used > together, otherwise it would be a dependency loop. (you could approximate > this by checking for direct dependencies only, which will cover all cases > we're likely to see in the near future).
Right. > > It's going to be even worse for the dep planning. I'm not sure how to > > solve it yet. > > So if you establish that base-3 and base-4 can coexist using the above > rule, can you then use that information when deciding whether two packages > can both be present in the plan? Yes, that's fine for checking if a proposed plan is valid. Adjusting the search algorithm to find such solutions is rather harder. Duncan _______________________________________________ Cvs-ghc mailing list [email protected] http://www.haskell.org/mailman/listinfo/cvs-ghc
