2020 OS Migration
project-trident.org
project-trident.org
Which is... fascinating. Switching out upstreams is one thing, but they're switching kernels in the process!
EDIT: more comments from finishing reading:
Looks like they do broadly intend to work with the Void upstream, so I'm quite pleased (as a Void user) that there will hopefully be more people contributing, especially on the musl branch.
> We will not be creating a migration path from the FreeBSD-based Project Trident to the new Void-based version. Please be prepared to copy your user data (/home/*) to a temporary location like a removable drive before installing the new version and copying the data back.
That's quite unfortunate. In particular, if they do integrate ZFS into Void including boot environments, there shouldn't be any reason they can't do it in place; just install into new filesystems and mount the old `tank/usr/home` as though nothing had happened. Unfortunately you would still lose direct binary compatibility, since Linux doesn't have a FreeBSD personality to use, but I don't see why you should need a new pool or home.
The true irony to me is that they probably could support it if they were going the other direction, since FreeBSD does have a Linux-compatible personality. Actually, you might still be able to leverage that going the other way... release X is based on FreeBSD, release X+1 uses the FreeBSD kernel but migrates to Void Linux userland complete with actual Linux ELF binaries and can still run FreeBSD binaries, release X+2 switches to just running the Void Linux binaries on a Linux kernel, and only moving one release at a time is supported. It'd just be an insane amount of work.
Switching kernel can actually be a pretty small thing compared to switching distribution! I remember using Debian/FreeBSD, Debian userland on top of a FreeBSD kernel. For most people, I bet switching away from BSD to GNU userland would be a bigger deal than moving from the FreeBSD to Linux kernel.
From the home page:
> Project Trident is a desktop-focused operating system based on TrueOS. It uses the Lumina desktop as well as a number of self-developed utilities to provide an easy-to-use system that both BSD beginners and advanced system administrators can feel comfortable running 24⁄7.
But after this change they are dropping BSD and TrueOS configuration tools. I'm curious how many users would stay through this transition (lots of other Linux OSes to move to instead).
Why not just say it?
"Sadly the graphics support is still lacking on FreeBSD since all the driver development is very linux specific, we therefore decided to switch kernels because graphics is a major part of the user experience."
The FreeBSD wireless driver space is also pretty dated. Only 802.11N is supported -- the one or two 802.11AC WNICs that have support just run in the slower N mode.
In short, the situation is getting better and isn't awful (especially on nvidia graphics and wired NIC hardware), but the concerns TrueOS has are totally valid and won't be completely eliminated even if FreeBSD stays caught up in the graphics space.
Having said that, if they use the same frameworks as the x86 drivers, the work on Intel and AMD ports should make porting them easier as well, since it all revolves around linuxkpi - shims that provide Linux-compatible API for kernel code.
Not only that, there's often some kind of dependency between GPU/display HW and platform clocks/IOMMU/etc blocks that we might have completely different drivers for. This embedded crap is super messy, why can't everyone use PCI for everything inside the SoC >_<
I dont think there is a direct relationship other than that, aside from ixsystems apparently also supplied some forum hosting for project-trident (more correctly: a subforum on their forum)?
Someone needs to get over themselves and start thinking about their users.