I hit the following error today: WARNING: One or more updates have been skipped due to a dependency conflict: dev-qt/qtgui:4 (dev-qt/qtgui-4.8.5::gentoo, ebuild scheduled for merge) conflicts with dev-qt/qtgui:4/4=[dbus] required by (net-p2p/transmission-2.80::gentoo, installed) (dev-qt/qtgui-4.8.5::gentoo, ebuild scheduled for merge) conflicts with ~dev-qt/qtgui-4.8.4[-aqua,-debug,glib,qt3support] required by (dev-qt/qtcore-4.8.4-r5::gentoo, installed) (dev-qt/qtgui-4.8.5::gentoo, ebuild scheduled for merge) conflicts with ~dev-qt/qtgui-4.8.4[accessibility,-aqua,-debug,qt3support] required by (dev-qt/qt3support-4.8.4::gentoo, installed) !!! The following update(s) have been skipped due to unsatisfied dependencies !!! triggered by backtracking: dev-qt/qttest:4 dev-qt/qtdeclarative:4 dev-qt/qtsvg:4 dev-qt/qtopengl:4 dev-qt/qtcore:4 dev-qt/qtmultimedia:4 dev-qt/qtwebkit:4 Looking at net-p2p/transmission-2.80, patrick updated the dependency to dev-qt/qtgui:4=[dbus(+)] after I had already installed it. It seems like portage was using the dependency from vdb instead of from the portage tree.
This is with sys-apps/portage-2.2.0_alpha188.
Also, re-installing net-p2p/transmission cleared up the error.
The thing is, slot-operator deps require that the atom from the installed package be used, since that's where the built-time slot/subslot is recorded. We may be able to handle it better if we merge the slot/subslot from the old atom into the new atom.
I guess the alternative would be to revbump whenever a dependency with a slot-operator changes.
We may be able to improve the backtracking logic to trigger a rebuild automatically, or add a --changed-deps option to rebuild all packages with changed deps.
not cool, re: useless revbump done for net-p2p/transmission referring this bug
(In reply to Samuli Suominen from comment #6) > not cool, re: useless revbump done for net-p2p/transmission referring this > bug I mean, I would have rather left any subslot operator ignored in the ebuild than get useless revbump forcing useless rebuild on users I see useless rebuilds as one of the biggest QA problems in gentoo-x86 lately
(In reply to Samuli Suominen from comment #7) > I mean, I would have rather left any subslot operator ignored in the ebuild > than get useless revbump forcing useless rebuild on users > I see useless rebuilds as one of the biggest QA problems in gentoo-x86 lately A revbump was the only way I am aware of to fix the above mentioned problem for users with net-p2p/tranmission already installed. Better a "useless" rebuild than an ugly message from portage telling you it cannot resolve dependencies.