No. And after reading through this thread, I must say you remind me of my hairdresser. Did you know that France would be a much better place if only the government dared to listen to his advice in matters of tax, employment, health and immigration policy? Only they don't, because they are afraid of the real break-through innovation. Hence the country is going down the drain. What a shame.

So now you imply that I have as much qualification to talk about systemd as your hairdresser presumably has about the management of France. But who — may I ask — is better qualified to comment about it than someone who has actually read through all the articles posted on Lennart Poettering's blog and the Fedora mailing list threads pertaining to systemd? Here most people on this thread are like most ordinary joes who went and voted 'yes' or 'no' to the E.U. constitution without having actually read it. Crowdism at it's best.

1 members found this post helpful.

Yes i know that they are optional dependencies, but the article is deliberately misleading. He claims a ton of features but he doesn't mention that if you want to have them you will need other libs too and not just the 2 dependencies. He presents facts the way it suits him (like the car commercials that say "up to 200HP" and "from 8000$")

*BSD,OpenIndiana,etc are OpenSource software so they matter.

Yeah, right. I encourage other OSes to share interfaces after I deliberately code in a way that isn't portable and violates standards. That way the code will be superior (as was pulseaudio )

Have you ever ran *BSD ? I run NetBSD and don't find it "in the stone age". Also FreeBSD with GPT+ZFS is a fine choice. Many linux users accuse windows users of having a "only windows exists who cares about linux" mentality but unfortunately many linux users have it too towards other OSes.

Without udisks, you won't have the dialog to automount a usb flash disk when plugged in, so you won't have "functionality that even Windows 95 had". In your case it is important to have all the functionality you can, but in *BSD users case who cares right ?

Anyway, i think i am going to take mRgOBLIN's advice and let the troll die. (/me goes to get fire arrows and memorize melf's acid arrow )

4 members found this post helpful.

I'm afraid I can't agree with KnutBluetooth being a troll - the BSD's, Solaris et al are being left behind _already_ in the desktop world - for example, in the Linux graphics stack world we have exactly the same situation, they've already been dumped in favour of kernel mode setting in the Linux kernel, which is a deliberate decision by the various maintainers - their view being why hold back Linux to keep the BSD's going - if they want new features, they can do the work themselves.

1 members found this post helpful.

In the same vein, my hairdressers' background includes having read all the relevant articles in the far-right press, as well as having attended lengthy symposiums in the local Café de la Poste. To close off this discussion, let me point you to a less famous quotation from Ludwig Wittgenstein: "Thereof one cannot speak, thereof one frequently goes ranting on and on at ball-breaking length" (Tractatus Logico-Philosophicus, first draft).

Well, I think that running systemd without any of it's optional dependencies is still way better than running sysvinit. Even with the minimum dependencies you still get simpler config files to replace shell scripts and the ability to track processes reliably.

So by virtue of being OpenSource they matter. Ok, then following this logic let's make every open source app portable to FreeDOS.

Sharing interfaces is not sharing code. One could (theoretically, as *BSD is missing much functionality in the kernel) code an init system tailored to *BSD which uses *BSD kernel stuff (such as kqueue for example) and which will implement the same interfaces as systemd. That is what matters to software that sits at a higher level in the userland stack. Just like Haiku/ReactOS are binary compatible and share the same interfaces as BeOS/Windows but they don't have the same code as the aforementioned OSes. Or as another example, if there ever is a Gallium 3D Direct X state tracker, will you expect that Window's implementation of DirectX be portable to UNIX ?

You ***can't*** code something that is functionally equivalent to systemd with only POSIX. So one might as well not use POSIX at all in this case. It's like saying that Linux drivers should follow POSIX so that they can be recompiled on *BSD. It just doesn't make sense.

Furthermore, Lennart Poettering said that people are welcome to use systemd's code as a basis for their own init systems. But he just won't maintain an #ifdef soup to accomodate every UNIX-flavor out there. And many of these flavors of UNIX simply do not have the kernel functionality required to create a systemd replacement to begin with.

Of course. And actually, for the record, I find the *BSD userland much nicer than Linux's. I use it as a server but I would never consider using it as a desktop OS right now as it's clearly far behind Linux in that area. In that area it's still in the stone age. And Linux itself in that perticular area, lives in the stone age compared to MacOSX, Windows or even BeOS. But redhat is doing something about it to catch up and one of those efforts is actually systemd.

It's a fine choice for a server system, yes.

Let's just say that if I wanted to run Windows apps, I wouldn't be running Linux. I do not expect Windows apps to be portable to Linux. I just expect to have equivalent apps on Linux and there in fact is.

Well, as I said, the *BSD developpers should care. Not the Linux ones. As long as Linux developpers cooperate so that interfaces are generic enough to suit everyone, I call it fair. But you should not expect Linux developers to code portable ***system*** components for all kernels out there or maintain an #ifdef soup (system is the keyword here). And btw, that's exactly the policy that OpenBSD people have been following too and much more uncompromisingly than in the Linux camp. I'm still waiting for the newer versions of openntpd to be portable. Oh wait, nevermind there's chrony now.

0 members found this post helpful.

I'd suggest you read a book called « Eristische Dialektik — Die Kunst Recht zu behalten » by Schopenhauer ^^

0 members found this post helpful.

I agree with that. I like Slackware's philosophy of keeping things simple and cautiousness about the "latest and greatest" stuff but I wouldn't criticize other distributions for trying out radical new ideas. Actually I'm thankful that they do, they are like the test mice in the lab. If there is any problem they are the ones to suffer, and we will later know which new software to select for inclusion in Slackware. And just like millions of people are still happy using Windows despite its bugs, there are many Linux users who wouldn't mind the potential bugs in the new stuff -- we Slackers aren't one of them, but those others outnumber us by far.

And while I would encourage cross-platform compatibility wherever possible, I wouldn't make it a requirement. Yes, BSDs are lagging behind in certain respects, and it would be unfair to delay the progress of Linux to help BSD keep up.

Unix/Linux and Windows were created with different priorities in mind: Security, stability in the former, and user experience in the latter. They both succeeded in realizing their main goals but they are both weak when it comes to their opponent's strength. The trend that I see in the last decade is that, the two sides are converging to each other; with Windows redesigning its security infrastructure (by reinventing Unix-ish ways) and Linux focusing on improved user experience. This is good for both sides.

Can't they do these improvements in Linux without abandoning the simplicity of the old designs, making people like the average Slacker happy? I believe there is no technical obstacle for that, but if no one cares or dares to write new software "the right way", then the answer is practically "no". In that case it doesn't make sense to resist and get stuck in the obsolete world, but the best one can do is what Slackware is already doing, that is, to wait until the new software is stable enough, and among the non-obsolete stable options, stick to the simplest one possible.

2 members found this post helpful.

As an administrator, I simply want a secure and stable foundation, preferably with no configuration helpers that could possibly mess with my configuration files. I know where the relevant stuff is, and I'm mostly using Vim and SVN for configuration. I'm also thankful when basic things like filesystem layout, system startup and the likes are somewhat perennial.

Now some of my users don't even know they work on a Linux system. They just type in their login and password, and then go on doing what they have to do on their PC: managing a public library, take appointments for the medical practice, print an invoice, whatever.

One of the basic principles of my company (http://www.microlinux.fr) is to always be thoughtful of this distinction.

1 members found this post helpful.

The problem are Windowze educated desktop designers. They suffer a lack of imaginations. For them W$ is always some kind reference point. Almost ideal desktop system (sorry maybe MacOSX is much better than W$). So my impression reading this thread is that systemd is at first place for desktop PC's. Rather boring.

@kikinovak:

You are right from your point of view, but let us not forget that the majority of computer users use these devices as personal desktop/laptops (or nowadays, variuous mobile devices) and they have neither the knowledge nor interest in configuring system files. They are just simple users, they like it when the system boots quickly, or the gadgets are automatically recognized when plugged in, etc. While I am not of that kind, I realize that most people are, so I can't blame the distros for responding to the popular demand.

2 members found this post helpful.

These users are probably better off with any of the "Linux for everyone" distros like Ubuntu, Mint, Pardus, PCLinuxOS and the likes. Some of these do a really great job in matters of automagic hardware support and configuration. My dad - an ex-manager for a worldwide electronics company - uses Ubuntu 10.04 on his laptop, and he's still singing praises since he installed it as replacement for Windows XP.

One example - among many - why I *do* favour Slackware. I'm running a public library management software on a LAMP stack, and the application needs a few exotic PHP modules, among which the Yaz module for the z39.50 protocol (the sort of things that makes public libraries exchange information about books). Now I've only seen two distros where php-yaz could be installed without any pain, and that's Debian and Slackware.

I totally understand you, after all, I'm using Slackware for similar reasons. I just wanted to point out that, while most of the new software may initially look "bad" by our standards, "we" are not big enough in number, or at least there aren't sufficiently many high-quality projects which "we" would appreciate and which could also compete with those others. The result is, anyone who doesn't want to get stuck in obsoletion is forced to go along with the majority after a while. I like Slackware because it "resists" the "bad" changes as much as possible. But (as Pat pointed out) we may end up having to use systemd someday. I wouldn't want that to happen soon, but instead of making sharp statements early on, it is good to have an open mind and keep an eye on how things progress.

Lennart certainly does, in the sense that he seems quite hostile to it .i.e. it's inefficient to consider the wider floss ecosystem when developing floss software. This may be true for Redhat, but projects like Debian have objectives that are wider and frankly more socially useful that "get rich using linux".

Setting aside the fact that ad hominem doesn't add anything to the discussion, this is also flatly false. As any visually impaired user left without a functioning screen reader thanks to pulse audio will attest, rapid adoption of new core technologies is a risky thing. The abstract technical case for systemd has been well made but I yet to see anything that addresses the wider risks to FLOSS that wide, rapid adoption presents.

4 members found this post helpful.

PulseAudio is portable as well as libcanberra and a slew of other programs he did. So that's factually incorrect. As I said before : it makes no sense for something like systemd to be portable.

Hurd and GNU/kFreeBSD are toy OSes. Nobody cares about them except for a handful of afficionados. And so to please this minority Linux's progress should therefore be hindered.

PulseAudio IS portable so I have no idea what you are trying to get at here since there is no obvious correlation between portability and rapid adoption of new core technologies.

Furthermore, before replying to messages you ought to read the whole thread because to quote myself:

0 members found this post helpful.