2007/6/19, ngaba@petra.hos.u-szeged.hu <ngaba@petra.hos.u-szeged.hu>:
I have the same opinion usually (see my earlier sortbydeps complaints .-). But now we just search for removable dependencies, if we miss some pathological cases this is not a big problem, the database won't be corrupted, just we don't remove a dependency which we could etc. Let me explain: I meant in my previous mail that we say that we cannot remove a package bacause it is needed by an other one without checking that it cannot be "replaced" by an other installed one (see alpm_checkdeps && multiple provision). But this "fault" is _very_ rare and this extra check would cause _notable_ slowdown, because can_remove_package is (/was?) called "often".
I agree that this function is much less critical, and that multiple provisions are pretty rare (is there any real case?). This is not a reason for not getting it right though. About notable slowdown, I don't know, are you sure about this ?