SoylentNews is people
SoylentNews
The good people over at Infoworld have published a story outlining why they feel systemd is a disaster.
Excerpt from Infoworld:
While systemd has succeeded in its original goals, it's not stopping there. systemd is becoming the Svchost of Linux—which I don't think most Linux folks want. You see, systemd is growing, like wildfire, well outside the bounds of enhancing the Linux boot experience. systemd wants to control most, if not all, of the fundamental functional aspects of a Linux system—from authentication to mounting shares to network configuration to syslog to cron. It wants to do so as essentially a monolithic entity that obscures what's happening behind the scenes.
This discussion has been archived.
No new comments can be posted.
systemd: Harbinger of the Linux Apocalypse.
|
Log In/Create an Account
| Top
| 98 comments
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
Tax reform means "Don't tax you, don't tax me, tax that fellow behind
the tree."
-- Russell Long
(Score: 3, Insightful) by Rich on Tuesday August 19 2014, @11:35PM
Nice clickbait article. Booting indeed needs attention. In the ancient days, a nice debian SysV init would scroll along, with lots of [OK], the odd [failed], and eventually the GUI would start. These days, on a current distro, all kinds of WTF-moments keep surprising me, mostly with silent failures indistinguishable from long delays when no DHCP lease is found or mount thinks an fsck is due.
Now here we go: The three much-discussed components (systemd, pulseaudio, and Wayland) are all born out of the idea to rip off what the Mac does with launchd, coreaudio, and WindowServer. (I just notice even to the point of the first letter capitalization.) The blueprints are followed to where the result is borderline bearable, quirks then get added to satisfy a Red Hat middle manager's pet peeve or just the developer's ego.
I don't follow the majority opinion with pulseaudio, it indeed seemed to expose a number of ALSA bugs, and these had to be fixed. The reasoning behind arbitrary PCM span lengths is also perfectly fine, where a lot of CPU cycling can be avoided on mobile. However, not considering Pro-Audio needs (and going for a unified solution with Jack) is where the mediocrity rears its ugly head. Also, from a system architectural view, putting certain audio drivers into the application layer mixer was outright braindead.
So, following the launchd concept isn't too bad either. The Mac works and it boots reasonably fast. However, mindless feature creep seems to ruin this one, again seemingly driven by Red Hat's day-to-day operational needs. I think such a boot system can hardly be reliably designed incrementally. It will need a through concept that implementers AND users understand and can agree upon. That also means seriously good documentation. There are fringe cases (e.g. Audio over IP, but IP over Audio (Softmodem): which comes first, if Audio is not layered or staged?) that have to be cleanly architected out, instead of including everything and the kitchen sink to cover all eventualities. I fear systemd is moving or has already moved past the point of sanity.
But is there a solution? Obviously Red Hat does what it does to make stockholders happy and community whining is considered as a slightly milder annoyance than a swat-evading-fly buzzing around in their offices. Shuttleworth has given up on fixing Bug 1 and only does mobile/cloud now. No more surprises like how well Dapper or Hardy worked compared to other distros of their time. Besides that, the Ubuntu experience past Hardy was also chasing the Mac like Red Hat, just with even more superficial engineering.
It would need a white knight out appearing out of nowhere, laying down a clean architecture AND amassing such a following that the wider community pushes Red Hat's momentum back. Which is not going to happen. So we probably have to get used to the new stuff as they give it to us or get into retro computing. But then, with retro, Firefox 2.0 still had a sane Look & Feel :)
(Score: 0) by Anonymous Coward on Wednesday August 20 2014, @06:50AM
I agree with most of your post, but I'd like to point out that I feel Wayland is exactly what must be done. I used to think the same way as you because "hey, everyone on the Internet opposes it, so it must be crap" but then I read this http://www.phoronix.com/scan.php?page=article&item=x_wayland_situation&num=1 [phoronix.com] and saw that presentation:
https://youtu.be/cQoQE_HDG8g [youtu.be]
TL;DR: most Wayland developers and supporters have been the X11/X.Org developers for many years and they know exactly what they're doing. See Wayland as what X12 should have been but never will.
It's a bit like OpenBSD guys rewriting OpenSSL into LibreSSL: they know what they're doing and they're doing it well.
The guys writing systemd seem to be doing just what Red Hat wants, not caring about consequences or crappy design.
(Score: 1) by Rich on Wednesday August 20 2014, @02:25PM
Bit of a late reply... I didn't elaborate on Wayland because I didn't want to stray too far off topic. I'm entirely with you here. I occasionally jest that X11 should be replaced with TCP/IP. The latter is much leaner yet provides roughly the same support for modern desktops. Wayland going for the WindowServer model (layered client bitmaps) is probably the universally most reasonable choice. Though if I had to architect a desktop myself, I'd go for the Be model with a policy bound application server and CarbonEvents-like network transparency at that layer, for a maximum of "teh snappy".
I think the most important "feature" of the new graphics stack is that drivers are separated from the window server. A window server is a high school project compared to getting accelerated GL on a modern GPU. With separate drivers, alternative approaches get a chance.