Does FreeBSD Have a Future?

Yes ... and the future is bright.

I have been using FreeBSD for almost two decades now (and still use it) both privately and professionally. I stared at 'dark' 5.x times with problematic SMP implementation and DragonflyBSD split.

These back then in 2004-2005 were the 'dark times' for FreeBSD.

Since then all things improved or been reworked.

Besides ZFS inclusion many great things happened in FreeBSD that are not available anywhere else like:

- GEOM Storage Framework

- Jails with VNET or Netgraph networking

- Bhyve virtualization/hypervisor written from scratch and ported to macOS and Illumos

- new LUA based loader(8) with support for ZFS Boot Environments - beadm(8)/bectl(8)

- TLS in kernel (Netflix)

- imported DTrace for better troubleshooting and looking for performance improvements

- Capsicum sandboxing Security framework

- 33000+ available packages in pkg(8) repository

- support for new/latest GPUs

- boot speed improved from 28 to 8 seconds: https://youtu.be/10XRCiBtyhA

- import of pf(8) OpenBSD firewall

- ported to ARM64 and POWER64 architectures

- move from M:N to 1:1 threading model in SMP

- new bsdinstall(8) installer and bsdconfig(8) companion

- sendmail(8) removed for dma(8) from DragonflyBSD

- ... and I definitely do not remember tons of other stuff (almost two decades)

Not to mention BSD license that gives you ABSOLUTE freedom and not just a glimpse of it.

Some of these features I described here:

- https://vermaden.wordpress.com/2020/09/07/quare-freebsd/

Regards.

> the only reason to select FreeBSD is the ability to close-source everything you're doing?

I doubt that's the reason, since you can certainly develop closed-source applications for Linux.

I'm moving to BSD specifically so I can stop using Linux, because I dislike the development direction that Linux is on.

Netcraft has been confirming the death of FreeBSD for decades now. It works and continues to work. It might not work forever, but why not use the (subjective) best tool while you can?

Linux is unlikely to implement a low churn kernel and userland; unless something comes up and the contributor count drops a lot. When I left a company that was kind of sort of stuck on running applications compiled for FreeBSD 4 to one running much more current versions, it was no big deal: everything worked better but the user experience is the same. Otoh, half of the Linux systems I use for work now use ip, the other use ifconfig, sometimes netstat, sometimes ss, etc.

As far as the road towards feature parity goes, it seems that Linux is mostly evolving technical solutions rather than replicating them. Dtrace on Linux is a big project being superseded by eBPF[0] - its the primary example that comes to mind. As a counterpoint, OpenBSD's pledge has been re-implemented on Linux using bpf & seccomp[1].

I still value BSDs as they are developed as complete operating systems rather than as cobbled-together distributions. Users benefit from a more cohesive experience on BSD vs. Linux.

>and the only reason to select FreeBSD is the ability to close-source everything you're doing?

I'd love to have a GPL-licensed BSD operating system. Ideally new features would be prototyped in BSD-licensed BSDs and eventually make it into a stable GPL-licensed version. (I know it sounds crazy...) I don't want to get my hopes up, but the people behind the Hyperbola BSD project have been working on it for some years now[2].

[0] https://news.ycombinator.com/item?id=22987696

[1] https://justine.lol/pledge/

[2] https://www.hyperbola.info/news/hyperbolabsd-is-an-operating...