>at least one statement in the thesis about Gentoo is no longer true (Gentoo prefix allows regular users to install and manage packages using portage on a varie

>at least one statement in the thesis about Gentoo is no longer true (Gentoo prefix allows regular users to install and manage packages using portage on a variety of systems [0]). I assume you're referring to the following sentence, but that's not what it's talking about. >As claimed in the introduction, deployment systems tend to have monolithic trust models. Typical Unix package management systems such as RPM [62], Debian APT, or Gentoo Linux [77] allow installation of software by the administrator only; software installed by individual users is not managed by those tools This is talking about a single instance of the package manager; for any given instance of the package manager (where an "instance" is defined by various things like the list of installed packages, package data, various package management databases, etc.), only the owner of that instance is allowed to install new packages. For a system-wide package manager, only root is (safely) allowed to install new packages. Even at the time, RPM and dpkg, and presumably portage, had the ability to create non-system-wide instances of the package manager, like Gentoo Prefix does; but even for such instances of the package manager, it's only safe for the owner to install new packages. (You wouldn't want random other users installing packages in your home directory!) Such instances don't share anything with the system-wide instance - it's essentially like creating a chroot or container. The reason this is being pointed out in the thesis is that Nix does not have this limitation; any user can safely install arbitrary packages into the system package manager, which are visible to the system, share package data, etc. I'm taking pains to be clear about this because, IMO, this is one of the places where Nix has the clearest advantage over traditional package managers.