many small programs do not need to be built for the whole system. There could be a user-emerge which would build a package into the users homedir
This might be better (more globally) solved by allowing portage to override the FHS conventions and define --prefix (which I'm hoping has already been submitted to Bugzilla or I'll be writing one ;-)
*** Bug 11040 has been marked as a duplicate of this bug. ***
I've given some thought to this as well. A *really* common mistake of people building packaging systems for Linux (and, for different reasons, for Unix) is ignoring the fact that there are people running multi-user boxes. On systems like this, having the ability for users to install packages in their home directories is, well, a requirement. Normally, non-admin folks do the tar/configure --prefix/make thing, but they then lose out on all of the package management aspects of really excellent package management systems like Portage. In my mind, user-mode Portage would work something like this: 1) check for permissions to do system-wide install. If no (or if a command-line option was given) install in user mode. 2) If in user mode, build a mini-Portage installation in the user's home directory. 3) Merge the global, system Portage DB with the user-mode DB to resolve dependancies 4) Install relative to the user's dir. There are some other issues to consider, but I think they're more incremental improvements than requirements for a first shot. For example, if/when a newer version of a package is installed globally, what do you do with the user-installed version? Similarly, what about when this happens to dependancies? Still, I believe that these can be resolved with POLS; for example, when the user does an emerge -pu world, if Portage detects a newer global version, it simply uninstalls the user-mode version. The biggest problem would be resolving how changes to the global system will affect user-mode installations. However, a user-mode Portage with problems like this would be better than no user-mode Portage at all. Thanks for listening.
Is this a dupe of pathspec...or..would you consider it seperate? Definately would be good to link to bug 87877. http://bugs.gentoo.org/show_bug.cgi?id=87877
Can go as a dupe. Nothing much happening here and the other bug covers it all. *** This bug has been marked as a duplicate of 87877 ***