Relying on udev to assign permissions to the virtgpu was very "proper"
but really excessive. We have a simple static configuration, built into
the kernel, so there are no dynamic shenanigans, we can always reliably
just chmod/chown the permissions. Let's go back to faster booting.
This project is Linux-specific (microVMs, libkrun, systemd, etc.), so
the flake now only exports packages and checks for x86_64-linux and
aarch64-linux systems. This prevents Darwin build failures.
Additionally, nixosConfigurations now set allowUnfree = true to allow
packages like zerotierone that have unfree licenses.
This adds a checks attribute to enable automated validation of all
build outputs. Checks are included for all packages, devShells, and
nixosConfigurations, with appropriate prefixes (package-, devShell-,
nixos-) for clarity and organization.
This extracts the testvm configuration into a shared nixosModules.testvm
module that can be reused across different architectures. The module is
now used by both testvm-x86_64 and testvm-aarch64 nixosConfigurations,
eliminating code duplication.
Additionally, nixosConfigurations have been moved to the top-level flake
outputs to follow Nix conventions, rather than being nested inside
eachDefaultSystem. The virtwl proxy reference now uses ${pkgs.system}
instead of hardcoded x86_64-linux to support both architectures.
