Guest post by It has been roughly a year and a half since the last release of the GNU Hurd operating system, so it may be of interest to some readers that GNU Hurd 0.6 has been released, along with GNU Mach 1.5 (the microkernel that Hurd runs on), and GNU MIG 1.5 (the Mach Interface Generator, which generates code to handle remote procedure calls). New features include procfs and random translators, cleanups and stylistic fixes, some of which came from static analysis, message dispatching improvements; integer hashing performance improvements, a split of the init server into a startup server and an init program based on System V init, and more.
Congrats on the integer hashing library performance improvements.
Hurd has been regarded as recurring comedy for a good while, but that’s better than tragedy, I wager
I prefer to think of it as the FSF’s incredibly insightful ongoing unintentional social experiment.
This some historical value (it’s difficult to find the old Mach code nowadays) but they really shouldn’t spend any time on this. The project is for all purposes dead, specially since rms kicked out the main architect.
As of lately the FSF only exists to relicense the legacy GPLv2 userland to GPLv3, but it doesn’t seem like the FSF even owns enough of the Hurd code to relicense it.
Perhaps GNU Hurd would start maturing after Stallman retires in the near future.
Exactly what happened?
Any news out of GNU Hurd HQ is always encouraging. Thanks guys.
I’ve been honestly looking forward to Hurd since forever. I found out about it at the same time I discovered Debian’s online repositories. A completely new foundation that wasn’t ready yet, but it’ll be great we promise.
It’s like a big, heavy, unwrapped Christmas present with blue shiny paper and a big red bow.
It’s going to come out and it’s going to be amazing.
It’ll be just like when I first found Linux distributions and they all came with thousands of software programs that did fancy computer things. .
The development of the Hurd is inextricably associated with nebulous positive qualities that no functional release will ever fulfill. I can’t say I want it to be released, but I do love reading these announcements.
Oh God, why?!?
Give me that over systemd anyday.
There’s a lot of init systems that are better than SysV init. Heck, pretty much everyone of them are, even systemd.
SysV is an abomination that belong in the far past.
systemd is an unnecessary evil, developed by people known for being entirely incompetent and utterly useless. sysvinit is lightyears better than systemd. If sysvinit is bad for you, you are doing it wrong. It is suboptimal for the silly bloated kitchen sink binary distributions (read Ubuntu and Fedora), but if you know what you are doing, nothing beats sysvinit (a pleasure to use with LFS).
If you need extra stuff for the sake of laziness there is always OpenRC.
And yes. The years have made me an old grumpy gaffer
And yet, the years didn’t teach you how to present an argument without trolling.
You may not like systemd and/or PA, you may think that they are plain wrong.
But the people behind it are anything *but* incompetent.
… And given the obvious fact (in your eyes) that “I’m doing it wrong”, care to share how many Linux system you manage? I’d love to learn how “to do it right”.
Edited 2015-04-18 10:37 UTC
Linus disagrees – and so do many others. systemd is for “click’n’point” administrators.
… Trollish remark. I’m not arguing with Linus, I’m arguing with you. Plus, as far as I know Linux didn’t call the systemd developers incompetent.
And let me quote him:
A *FAR* cry from your trollish remark. (And Linux isn’t known to taking a subtle approach)
Which brings me to the my original question:
I’m a patient man.
Edited 2015-04-18 13:23 UTC
I’m not the one trolling. There is no trolling in my posts. They are all genuinely meant. They may rub you the wrong way, but disagreement isn’t trolling. No matter how you feel about it. And repeating a wrong statement does not make it more correct. No matter how often it is repeated.
The number of linux systems one maintains is irrelevant to whether or not it is “done right”. Quantity is not quality. But the answer is five. Again the number does not matter; also one should count in other persons maintaining similar systems and their experiences with similar systems. If you look at LFS and gentoo you’ll see that PulseAudio and systemd are strongly disliked – and for good reasons.
Lets reiterate the facts: (And please note that I do not argue whether systemd is good or bad, I doubt that we’ll agree)
1. You called the systemd developers incompetent (as opposed to “I disagree with their design decision” and/or “they are doing it wrong”) and than (wrongly) used Linus to justify the original claim.
2. You claimed that anyone that requires systemd “is doing it wrong”, even though you’ve failed, repeatedly, to display any type of qualifications that somehow qualifies you as someone that “does it right”.
If this is not trollish behaviour, I fail to see what is.
Edited 2015-04-18 17:50 UTC
Nope, SysV init is terrible. Always was, always will be.Yes, systemd is horrible in many ways but it does manage services right.
Upstart, runit, daemontools etc etc, all of them are superior to SysV init and the terrible, terrible init scripts and runlevels. If the managed services are not run in the foreground you should go back and redo your design. It’s as simple as that.
systemd only manages to do everything wrong. But we obviously see it very differently. Apart from simpler configuration files I see no improvement in systemd. From my perspective I only see glaring mistakes and a horribly architecture that makes sysvinit look like morning dew in the desert. YMMV.
Any specific examples you can come up with without googling for answers? I’m just asking because I suspect you don’t know what you’re talking about. I mean, since you fail to construct an argument and all.
Now, watch that one ^^
That’s actual trolling.
To “No it isnt”. Let’s just start with the first glaring mistake, that of not sticking to one thing and doing that one thing well. You may not see it as a glaring mistake, but I do.
EDIT: It is interesting how much a religion systemd has become for some users. The slightest critique and the fanboys come out. It is almost apple-ish.
Edited 2015-04-18 23:38 UTC
It’s not trolling. I’m asking one specific person a specific question, demanding technical answers. You’ve failed to give one. Like I said, you’ve got no fucking clue.
And re: religion: you’re projecting.
1) You don’t get to demand anything.
2) You got a technical reason for my dislike of systemd, related to design philosophy of architecture. There are other issues, but I considered it enough with one reason. That’s all you asked for and you got it.
3) Yeah I am. But that doesn’t make it wrong. Nor does it make it inherently right. Your attitude indicates I’m right. The heavy modding down of entirely non-aggressive posts also indicates I’m right. Modding down of non-spam posts only occurs during religious wars.
Edited 2015-04-19 12:38 UTC
There’s no way my attitude would indicate anything about you being right or wrong. Sorry, but you seem to be something of an idiot.
Since you have relied on ad hominem attacks from the very beginning and getting mightily upset about valid technical criticism of software, I’d say you’ve made my point.
No, you don’t know what an ad hominem attack is either.
As a Linux user – home, a couple of small servers and 40 odd work stations I^aEURTMm quite interested in this argument, but admit to being painfully ignorant of the real issues, I also admit to not using, or likely to use Gentoo or LFS but more likely to use Ubuntu, Arch, Cent or Debian.
Dylansmrjones your argument seems to be systemd is bad because it is monolithic and does too much – it is not the UNIX way, do one thing and do it well (which you see as the right way it do things). SysV init, however, does things the ^aEURoeright^aEUR Unix way i.e. 1 thing and well, therefore if it isn^aEURTMt broke don^aEURTMt fix it.
Soulbender you argument seems to be SysV is bad because it is fundamentally broken, it does not do 1 thing well (the UNIX way), but does one thing badly, for example it allows managed services to be run in back ground rather than foreground. Systemd is an improvement for example running services more appropriately. You are less interested in ^aEURoethe UNIX way^aEUR and see some real advantages in having a more comprehensive init system, that allows for more over arching administration of the system. You don^aEURTMt see ^aEURoethe UNIX way^aEUR as being necessarily the most appropriate way of running a system.
Have I characterised these opinions correctly? If so let^aEURTMs avoid being trollish accusing each other of being trolls and make the argument as I am genuinely interested in the answers.
Not quite.
With the exception of systemd, all the init systems that I mentioned are more “the UNIX way” than SysV init. They are simpler, easier to configure and works better.
Unfortunately it seems that we’re stuck with SystemD as the new standard. Thanks a bunch, Red Hat…
Well, for one Soulbender was not the one to call me troll. It was “gilboa” and “no it isn’t”. It is them who cannot handle dissenting views on the greatness of systemd.
TODO:
*insert joke on Soulbender, increased age, grumpiness and hiding* :p
Almost everything, yes.
As you already said; easier configuration files. I don’t have to write a damn shell script to run a service.
Services can be reliably managed by running in the foreground. No retarded PID files and backgrounding.
Dependencies so that you can be sure services are started in the right order rather than, as with SysV, hoping that your ordering works out.
Of course, being SystemD they had to f–k it up with cgroups and other pointless stuff. I guess Red Hat needed it to be “enterprisey”…..
Edited 2015-04-20 03:25 UTC
I can understand that, even appreciate it. But it is not enough for me to compensate for the nastiness in the rest of it. I’d like to see those configuration files with sysvinit though :p – possibly in combination with openRC.
That’s one of my biggest beef with systemd, apart from being big, bloated and non-unixy – and the binary logs. The “enterprisey” shit coming with it.
Depending on taste and needs one may however choose one shit over the other shit.
Give me sysvinit with openRC and the easily readable configuration files from systemd. Of course, with LFS (and to some extent gentoo) you never have so many services running that it makes ordering difficult.
Whatever happened to initng? It looked so promising.
Oh God yes, don’t even get me started on that. It’s up there with GNOME’s binary config files as far as idiotic ideas go.
I was always a fan of runit/daemontools and later Upstart but even Ubuntu has now capitulated to SystemD I hear
Because it works and maybe perhaps the people involved with Hurd aren’t really worried about a few microseconds saved on system startup?
I don’t know systemd but over the years (approx 30) I’ve gotten to pretty well grep SystemV startup and how it works.
At the moment I’m not sure if I can be half-arsed to learn ‘systemd’. I have a Meh attitude towards it. If it works then great. If it stays out of my way then great. At the moment I’m not doing anything in anger on any systemd enabled kernels so my attitude may change when I move to it full time.
The problem with SysV init is not that it’s slow, it’s that it is fundamentally broken and unnecessarily complex. Symlinks everywhere and a bunch of runlevels that absolutely NO ONE cares about any more. The only “runlevels” anyone need is normal boot and rescue.
Don’t even get me started on how idiotic up it is to store the PID in files on the filesystem and put programs in the background.
The work on sysv started when it was the only supported system, but more important, if I understand it correctly, most of the work is groundwork (and requirement) to support any proper init system. Check out the reports from the 2013 GSoC which implemented this, starting from an initial draft:
https://teythoon.cryptobitch.de//categories/gsoc.html
The more I think about it, the more I think that the Hurd is the best example of a failed software project that could ever exist.
Not because it has failed, but because it CAN’T fail. Nobody can buy out the Hurd. Nobody can pull the plug on the Hurd. Just enough people care about the Hurd to keep it going forever.
The Hurd has been crystallized as a museum piece of all the things that can go wrong with an ambitious project.
It has been a delight to watch the development of the Hurd and it is my sincerest hope that the project continues development indefinitely with ever-expanding requirements that keep it from being deployment-ready until the end of time.
Mandatory XKCD link: http://www.xkcd.com/1508/
Can someone enlighten me about what’s so wrong about HERD?
Stallman…
Still doesn’t explain what is so bad about _HURD_.