On Friday 09 June 2006 14:04, Stefan Schweizer wrote: > Please, do not assume our users being stupid. They know that they are using > an ebuild from the sunrise overlay with zero support. They deliberately > typed
You have said stupid, not me. Some won't care enough, I'm quite sure about that. We had such invalid bug reports occasionally in the past and I expect this to happen more often, the easier and more common dealing with overlays becomes. Regarding "zero support": Making this abslutely clear is what I miss looking at overlays.g.o. > "svn co http://overlays.gentoo.org/svn/proj/sunrise/category/application" > "emerge application" > > And also there are only applications from maintainer-wanted or > maintainer-needed allowed in the overlay. Because packages are not supposed > to overwrite files from other ebuilds it is unlikely that they can cause > any damage to applications that have not been directly installed from the > overlay. maintainer-needed is imho not acceptable at all, as any dev trying to clean up bugs, won't know if a bug report comes from a user of the main tree ebuild or from your overlay. > > Also some warning that an overlay may > > break the tree or fubar the users system > > That is not the intention of the overlay. If it were intended, it would be malicious. Even if not intended, it doesn't mean tree breakages won't happen. Some dev may change an eclass, without taking overlay ebuilds into account (and he doesn't have to), but the change may break all ebuilds inheriting the eclass in an overlay, leaving all the users of the overlay with a broken tree. And to make that clear: Eclasses in overlays are only "sort of" acceptable, when the same team handles the eclass in the the main tree, as eclasses in overlays hide the main tree eclasses. Carsten
pgpU7l3V10Wea.pgp
Description: PGP signature