jmtd → log → FLOS ≠ UNIX
I just read this excellent blog post called "Linux Future" which defines a term "FLOS" to describe what the current direction of Linux development appears to be. Specifically, it addresses the divergence from what one might call the "UNIX" philosophy in new core technologies like systemd.
I think sticking a new name to define this philosophical direction is a great idea. "FLOS" is something that both proponents and haters can rally around. It acknowledges the move away from the UNIX way of doing things, and doesn't overload a term like Linux, "ownership" of which is claimed by both camps (and others who just want to keep their heads down and get on with things!)
The next step, after embracing it, would be to define the strategy and the roadmap. Where is FLOS going? What's the endgame? Perhaps with such a vision we may end up with some technologies which will stand the test of time, like Apple's CoreAudio (10 years old) or Microsoft's win32 API (now 20 years old), rather than the ludicrously high plumbing turnover we endure right now.
(Oh my goodness, I finally wrote an actual Linux blog post.)
Comments
Sorry, we do not do FLOS in Debian. Debian is the Universal OS, which includes existence outside of Linux. And FLOS.
And there will be no /usr move, nor a systemd-only world, in Debian. I’ll GR if necessary.
@Anonymous: thanks for your comment. I didn't say anything about a /usr move or forced systemd-only in Debian. The latter is of course an entirely different thing from the option of systemd in Debian (what we have now) or a default of systemd in Debian. All different.
I doubt you will need to consider a GR, if things ever reached the stage where a group of developers were organised enough that systemd as default was viable (they would not propose it as mandatory) the momentum would be such that I doubt you'd win that GR. But I doubt we'll ever get there. Indecision and inaction reign supreme after all. The status quo will stay that way.
What Debian does nowadays is sadly rather irrelevant in the wider Linux world. Ubuntu has proven that you can build on the solid base of Debian and still diverge quite seriously in key places such as the init system. Everyone else has gone down the systemd road, and as time goes on these FLOS folks will continue to diverge from Debian (or visa-versa) until they barely resemble each other anymore. I guess Debian now aligns more with UNIX than FLOS, and I guess Debian in future will continue to do so.
I was once a firm believer in "universal os" myself (despite it not being part of the social contract, merely a tagline) but in recent times have developed a new take on it. Perhaps I'll write about that somewhere.
It's sad that the need to replace the init system with something sane and reliable is now clouded by the urge of systemd to include more kitchen sinks. systemd's init part is pretty and much more sensible and usable than for instance upstart's (and I'm coming from runit here). And the maintainer of the non-Linux ports already said that we should tell them what Linux's going to use and they'll write a converter to sysvinit. Even that would be easy with systemd's declarative unit files. Technically there's so much wrong with sysvinit and its missing supervision and also with the various inetds that only have half-hearted support for, e.g., IPv6.
I agree with about the upper half of the linked blog post, but it just degenerates to the usual rant below that. No surprise that the comments section just attracts the same ranting folks you always get when discussing the merits of systemd.
@Anonymous:
0
↑ this is the amout of fuck that people actually working on Debian on desktops give to such comments.