(When using USE="vhosts",) emerge --depclean will suggest unmerging older versions of web applications even if they're installed to locations using webapp-config. I'd like to have depclean not suggest webapps for uninstallation if they have installations via webapp-config. Perhaps this could be a FEATURES option.
Perhaps depclean is flawed by design and you shouldn't use it as such. Perhaps depclean should imply --pretend because of the minor number of cases where it actually yields exactly what you wanted. Perhaps the Portage developers can explain in more detail why the package manager cannot or should not make exemptions based on conditions beyond its control (i.e. after merging and before unmerging. :)
And another one... Perhaps webapp-config is flawed.
I find that in most cases declean does exactly what I want it to (and I have --ask in my default emerge options, so I always get to check the list). This is just a suggested feature to, in my view, make it work even better. While if the issue reported in my other bug (#232333) is intended, I would consider this a very highly desired feature, even if #232333 isn't intentional and is fixed, I would still like this feature.
Perhaps this is a variant of bug 198200. The recommended practice is to run `emerge --noreplace =cpv` to have an atom for a specific slot added to the world file so that the package won't be removed by depclean.
(In reply to comment #2) > And another one... Perhaps webapp-config is flawed. > To install multiple revisions of ebuilds side by side to the degree webapps SLOT=$PVR implies or toolchains multislot "feature" is indeed conceptually broken, misuse of the package manager as sidestepping any security efforts. However, as always it's matter of perception. That --depclean doesn't see package:slot (how about coining this e.g. "sub-package"?) as individual entities is indeed a bug,though. And this will hit lots of innocent users when KDE 4 gets unmasked, so it would be nice to ses a fixed Portage version soon.
As far as I can see this is a duplicate of bug 198200. If this is not the case then please reopen with an explanation. *** This bug has been marked as a duplicate of bug 198200 ***