vimanuelt There's that "re-inventing the wheel" thing again, that BSD is accused of. I usually answer by referencing the "it's not just the kernel" argument. To run a true BSD desktop, we really would have to reinvent the wheel, wouldn't we?

      Robin
      If you are referring to a monolithic BSD-native desktop environment comparable to the Oberon desktop operating within the Oberon system then yes absolutely. Such a design would constitute a tightly integrated interface where the desktop is not a superficial layer placed upon the operating system but an intrinsic component of its overall structure. For FreeBSD this would represent a substantial departure from current conventions. It would entail a purpose-built graphical environment developed in alignment with BSD design principles rather than as a retrofitted adaptation of Linux-based models.

      The adoption of an Oberon-like approach would promote architectural clarity and functional unity. Rather than assembling a system from disjointed components such as display servers window managers and session protocols it would provide a singular coherent environment where the graphical interface serves as a natural extension of the operating system itself. This would likely result in improved performance reduced complexity and a user experience that is both distinctive and inherently consistent with the goals of the FreeBSD project.

      If on the other hand you are referring to a desktop environment intended for use across both Linux and BSD systems then no. Such environments inevitably reflect the priorities and structural assumptions of Linux development and they tend to introduce dependencies and abstractions that conflict with the design ethos of BSD.

          I'm all in favor of a DE that is native to BSD rather than "borrowing" from Linux. But we agree that in a certain sense at least, that requires a re-invented wheel custom made for BSD without the Linuxy cruft as a workaround.

          I see it more as a lesson in achieving success. The most prominent desktop operating systems, such as macOS and Windows, adopt a single, unified desktop environment. This consistency enhances usability, simplifies support, and enables deep integration between software and system design. It is a model that works not because it restricts choice, but because it prioritizes coherence.

          For BSD, the answer is not to imitate the endless variety of Linux desktops, but to bring the desktop wars to a close by creating a native, cohesive desktop designed specifically for the platform. This would not only reduce duplication and improve the user experience, but also foster a distinct identity for the BSD desktop. A clear and consistent interface signals maturity, stability, and a unified vision.

          The economic advantages are evident. Reduced complexity results in lower support and documentation costs, fewer user-facing inconsistencies, and a system that can be confidently presented to users, developers, and vendors. A BSD-native desktop is not about limiting freedom. It is about directing effort where it offers the greatest long-term value. It is also about affirming that BSD is not merely a technical foundation, but a complete and self-defined computing environment."

          vimanuelt … This could be added to the about or elsewhere. …

          grahamperrin don't make the About page longer than it already is).

              vimanuelt the "Yes, …" for GhostBSD hardware compatibility is overoptimistic. It's based on FreeBSD, which you describe as variable.

                  grahamperrin
                  That distinction is based on default configuration and user experience, not the underlying capabilities of the FreeBSD base system. Here's a breakdown:

                  GhostBSD: Good

                  • Preconfigured for desktops and laptops: GhostBSD includes additional drivers and firmware (e.g., drm-kmod, Wi-Fi firmware, and HAL) by default.

                  • Automatic configuration: The system attempts to auto-detect and enable hardware like graphics, touchpads, and wireless adapters during installation.

                  • Ready-to-use GUI: The user lands in a working desktop environment with graphical resolution and input devices configured.

                  • Extra tooling: Includes sysutils likecups pre-enabled for things like printing. Xconfig is used to setup X11.


                  FreeBSD: Variable (Moderate)

                  • Manual setup expected: Users must explicitly install drm-kmod, firmware packages, and sometimes even write custom rc.conf and loader.conf entries.

                  • Bare-bones installation: No desktop environment or graphical login is included by default. A functioning desktop requires manual setup.

                  • Better suited to servers: The defaults assume headless use or custom configuration, especially for hardware acceleration and desktop interactivity.

                    Here is an amended table. :-)

                    grahamperrin

                    The earlier phrase "manual setup required" referred to user effort during installation, especially for desktop or non-standard configurations.

                    ⚙️ Details:

                    • FreeBSD installer (bsdinstall) does support ZFS out of the box, but:

                      • The user must explicitly choose ZFS during installation.
                      • There are multiple options (e.g. single-disk, mirror, raidz) that require familiarity with ZFS terminology.
                      • Advanced features like boot environments, snapshots, and rollback require extra setup post-install (e.g. using beadm or zfsbootmenu).

                    • GhostBSD, by contrast:

                      • Uses ZFS by default, with boot environments automatically created.
                      • Integrates a GUI snapshot manager (timeshift-like functionality).
                      • Offers a friendlier experience for users who may not know the internals of ZFS.

                    • Ubuntu:

                      • Offers ZFS as an experimental install option on root.
                      • It is not the default, and some tools (like Ubiquity or Subiquity) may not fully support complex ZFS setups.

                    🟢 So why change it to just “Yes” now?

                    Because the question of support is different from the question of user experience.

                    • Technically, FreeBSD supports ZFS fully and natively.
                    • Describing it as “manual setup required” is misleading if interpreted as a lack of capability.
                    • To avoid implying FreeBSD is deficient in ZFS support, it is more accurate to simply state "Yes", and clarify elsewhere if needed that ease-of-use differs.

                        vimanuelt The user must explicitly choose ZFS during installation.

                        Not really.

                        ZFS is the default.

                        A user with technical knowledge might prefer to explicitly choose UFS.

                          vimanuelt FreeBSD installer (bsdinstall) … extra setup post-install (e.g. using beadm or zfsbootmenu).

                          ZFSBootMenu is for Linux, "… Originally inspired by the FreeBSD bootloader, …". I'm not aware of a port to FreeBSD.

                          FreeBSD

                          bectl is in FreeBSD base. bectl(8)

                          beadm can be added from FreeBSD ports collection, sysutils/beadm. beadm(8)

                          Multiplatform

                          jmdavis/bemgr – manage ZFS boot environments on FreeBSD and Linux.

                          Linux

                          johnramsden/zectl | zectl ZFS Boot Environment Manager for Linux · John Ramsden (2020)

                          ZFSBootMenu – distribution agnostic. Probably better-known than zectl.

                          Please note, the list above is not intended to be comprehensive.

                          The README for bemgr was recently expanded, it now includes:

                            vimanuelt Ubuntu:
                            Offers ZFS as an experimental install option on root.

                            It's true that root-on-ZFS is experimental, however I'm not aware of any significant bug.

                            At https://bugs.launchpad.net/ubuntu I can't find an easy way to perform a refined search. For "root on ZFS", the top match is a 2016 report that is not related to root-on-ZFS; and so on.

                            A couple of weeks ago I found this: Zfs with encryption install doesn't boot after install from current Ubuntu 25.04 ISO - Support and Help - Ubuntu Community Hub. Not reproducible by me.

                                grahamperrin
                                No, it is accurate. We are referring to the default installation. Neither FreeBSD nor GhostBSD provides a development-ready environment out of the box.

                                    grahamperrin
                                    During installation, Canonical labels ZFS support in Ubuntu as experimental. If you wish to learn more about the specific issues, you may contact them directly. It is likely that their tooling around ZFS remains in an experimental state, which justifies the label. From their perspective, this classification is fair; they are considering the experience of end users rather than that of developers. The definition of “done” often varies: users expect reliability and polish, not just functional implementation.

                                        vimanuelt It is likely that their tooling around ZFS remains in an experimental state,

                                        I doubt it.

                                            vimanuelt … it is accurate. We are referring to the default installation. …

                                            You state "No" for FreeBSD, where the tools can be added.

                                            vimanuelt

                                            You state "Yes" for Ubuntu, where the tools can be added.

                                            If all three columns (for the three systems) become either "Yes" or "No", then the Developer Tools row does not belong in a table that describes uniqueness.

                                                  grahamperrin
                                                  That is true. Also, most users may not even care about developer tools.