Hi, In virtual/jdk/: jdk-1.4.2.ebuild rdepends on =dev-java/apple-jdk-bin-1.4.2* jdk-1.5.0.ebuild rdepends on =dev-java/apple-jdk-bin-1.5.0* jdk-1.6.0.ebuild rdepends on =dev-java/apple-jdk-bin-1.6.0* On my system with today rsync (Sun, 24 Jan 2010 10:00:01 +0000) there are no such ebuild in dev-java/ Actually there's no apple-jdk-bin at all. It cause problem with emerge: #v+ emerge: there are no ebuilds to satisfy "=dev-java/apple-jdk-bin-1.5.0*". (dependency required by "virtual/jdk-1.5.0" [ebuild]) #v-
uhm, oops, sorry
*** Bug 301479 has been marked as a duplicate of this bug. ***
apple-jdk-bin exists in some prefix overlay AFAIK. Since there are many alternative deps in the jdk virtual, it's simpler for maintenance to keep one copy in tree and not in the overlays as well. The problem hoewer is that some of the other alternatives (sun-jdk etc) are masked by license, and portage apparently won't tell you that when the first alternative it encounters is a missing package. I've therefore moved apple-jdk-bin down the list, and in my tests it now reports a missing license of sun-jdk which is hopefully instructive for the user :) I've tested with 2.2 so hopefully 2.1.7 branch will do the same. Please test. CCing portage to comment and see if they want to prefer reporting masked licenses for later alternatives even if the first alternative is a missing package.
ok, I've added the package in gx86, as it shouldn't have been committed (the virtual) like this in the first place. So sorry about that.
(In reply to comment #4) > ok, I've added the package in gx86, as it shouldn't have been committed (the > virtual) like this in the first place. So sorry about that. > We can close this then. I've reopened bug 288083 for the IMHO suboptimal portage reporting.