Hacker News Reader

That's why Ubuntu became so popular, and Mint after it. There was a cohesive vision that a Linux OS can be more than a collection of text processing utilities. That's also why Ubuntu is so reviled in some Linux circles, because they dared to make a choice for their users, dared to try something new, dared to implement something that wasn't a direct continuation of decades-old software. Once they proved it was possible, others started jumping on board, giving Linux the freshest coat of paint it had ever seen. It was a serious contender for the year of the Linux desktop, especially with the rise of netbooks.

But then politics happened. If it doesn't work the same way it did in 1968, it's not allowed in the OS that was the child of the 90's. Pulse Audio, Unity, Gnome 3, systemd (along with anything else Poettering has touched) are evil because they're new. Unix never did it, so Linux can't.

GNU's Not Unix (yet... we'll get there someday though!).

"That's also why Ubuntu is so reviled in some Linux circles..."

I sometimes wonder if good visual and interaction design can succeed in the open source model. In software development, tasks can be parcelled out in a modular way to many developers. But this model doesn't work well for visual and interaction design. If you want all the parts of an Operating System or piece of software to feel consistent and cohesive, you need a more focused approach. And that means handing design tasks to a designated design team and entrusting them to make the UX decisions. But with open source projects, we all want a say in the UX (me included); thus the endless (and often bitter) discussions about how a piece of software should behave.

The model of open source code has influenced and inspired countless companies. But who can point to examples of visual and interaction design from open source software that has had a similar effect? I see the influence going much more the other way i.e. open source software often follows ideas or conventions from commercial companies. I'd love to see this change. But how?

In my opinion, parcelling out software development in a modular way doesn't work that great neither if you don't have guidelines, api definitions – community enforced styles of doing things.

I think, the same thing works with visual and interaction design. But you'd need a small group of people to make the basic UX decisions and write concise guidelines others can follow. Like Apples «Human Interface Guidelines»[0].

This isn't really purely democratic, but I think it's almost impossible to find the "one way of doing things" if everyone has their say in every micro decision.

For me its much more important to have consistent user experience than the details of it. Its like tab vs spaces in programming or semicolon vs no semicolon in Javascript. I have my personal preferences, but i'm fine with either one as long as it is consistent at least within the project.

[0] https://developer.apple.com/library/mac/Documentation/UserEx...

I would suggest it is democratic - there is and always has been a tension between our elected officials bein our representatives and our delegates.

The same goes for any democratically appointed UX team.

Yes your right. But i'm swiss, so a lot of countries dont really feel democratic to me:) Not enough power for the people – too much for those elected.

On the other hand, we, the people are soon to vote whether Switzerland should terminate european human rights contracts. It's insanely democratic and it scares the shit out of me because it might actually get a majority.

You can shoot yourself in the foot with too much democracy.

I thought the UK was the only one insane enough to think human rights were a pick and mix. I recommend Conor Gearty on why the Uk should not pull out (LSE podcasts on iTunes)

We're not directly voting against it. The referendum is carefully titled "Swiss law before foreign law" which would include the human right convention among every international law. The Advocates emotionally argue that the swiss should never be dictated by foreign judges, we can perfectly handle everything ourselves. Unfortunately a lot of people in Switzerland live on ivory towers, thinking their better than everybody else on the planet and agree with that statement.

The right wing fear machinery fed with millions of donation money has really poisoned the political climate over the last decade. Whether you agree with their points, a system like ours, where basically every citizen can trigger a nation wide vote about almost anything (e.g. reintroduction of death penalty) only works if the debates are driven by reason, not populism. The other parties recently jumped on the same train, saying it's the only way to defend against those scare mongers … sigh

I still love our system, it gives the people not only power but also a sense of responsibility. It's not the government that fucked things up – you did. But it doesn't make a country completely immune to dedicated actors with enough influence and money like many believe.

The modern marketing machine has indeed done a number on democracy. And with marketing being related to money, and money invariably congregating on the right, well...

> Pulse Audio, Unity, Gnome 3, systemd (along with anything else Poettering has touched) are evil because they're new.

At least for Unity and Gnome 3 (including GTK3) I'd rather say: Great ideas but they're evil because they are badly designed, full of bugs that don't get fixed due to complexity, constant API changes, almost no documentation or only wrong and outdated documentation and a lack of flexibility that annoys even the casual Desktop user.

Nobody is against progress - most people love to use OS X as a UNIX GUI these days... it's not perfect but it's not that bug ridden mess GNOME became.

>Nobody is against progress

You'd be surprised. Just listen to the arguments against giving Linux a more logical directory setup (somewhat like OS X's).

Change is not the same as progress. A lot of people are against change, for reasons both good and bad. I'm not familiar with the arguments over directory setup, but I'd almost bet money on them centering around tradition.

Personally, I think the various flavors of directory setup are already so divergent that tradition no longer holds massive value, but that's not something to be proud of.

How is OSX's directory setup more logical than the traditional Unix one? Be careful not to confuse your familiarity with one system with it's inherent qualities. I have switched between BSD an Linux as my main desktop OS a few times and had this "damn, this is so much clearer in x" feeling in both directions.

> Great ideas but ...

The problem is the comparison point.

When original Gtk/Gnome came out, it was a WAY worse BPOS than Unity or Gnome 3 are today.

The trick was that when Gtk/Gnome came out there was no existing solution. So, even a bad one was okay.

The problem is that the new solution is competing against an existing solution that has more than a decade of development behind it.

It would take a Herculean effort to win that comparison.

> The trick was that when Gtk/Gnome came out there was no existing solution. So, even a bad one was okay.

There were existing solutions, e.g. Motif or Athena widgets if you wanted a GUI library, or FVWM if you wanted a desktop environment. Of course, GTK/GNOME was a significant improvement.

Mogrief was a non-starter due to license. Lesstif was a pile of fail forever. Athena widgets? Are you trolling me? See Xfig for an example of how bad those are.

Qt and Gtk were a breath of fresh air. And Qt got everybody's dander up due to 1) license and 2) C++ requirement.

So, Gtk basically stepped into a vacuum.

Perhaps the reason pulse and gnome were disliked has something to do with distros that don't support the old versions after a new one comes out and foisted unfinished buggy releases on their users before either was ready for prime time?

You need to take a few steps back to fully understand the so called systemd controversy.

1) A number major commercial UNIX vendors have moved in the direction dumping the old startup scripts in favor of something more explicit and declarative. This isn't just engineering for engineering's sake, major corporations have customers with needs. Redhat and company are no different.

2) It's incredibly easy to be vocal. What's really lacking is a compelling alternative. The most vocal opposition have made anonymous web sites, have made tons of noise but done very very little in the way of actually solving the problem, many even refuse to acknowledge them.

It's really easy to think it's a civil war of sorts but the reality is it's mostly just a very vocal group that are against it, and FWIW, they don't seem to contribute that much. The BSD world has had it's share of these too, it's just a more dramatic fork when you startup OpenBSD project or NetBSD or DragonFly or one of the more minor forks.

Openness means you get to see the bad as well as the good, the decision. A very vocal group of folks were crapping on GNOME 3 when it came out, now it's good. If you think systemd was contentious, while I love the idea of unifying configurations, just wait until that debate starts... It's probably not possible to represent all of the different things that the "tower of bable" has without a nightmarish XML schema, if it can be done. I welcome the ideas though, especially if FreeBSD get's serious about mobile and power, there will certainly be some interesting stuff.

Explicit, declarative? Sounds nothing like what i have seen so far.

There are already several alternatives to systemd, starting with the systems that systemd is replacing that have worked well for years.

That comment sounds like you either don't know what systemd does or are just trolling. If you want to have a constructive conversation, talk about how something which is tedious in e.g. SysV init is better solved by OpenRC/Upstart/launchd/SMF/etc. rather than systemd. (Repeat for any other aspect of interest)

How can he possibly be trolling? He's pointing out that the existing systems work fine and have for years, and that's the full extent of his post. This is why people hate systemd: the hordes of people who do personal attacks on anyone who even disagrees with them. You're doing it right here: anyone who disagrees with you is a troll.

> He's pointing out that the existing systems work fine and have for years, and that's the full extent of his post

That was my point: there are well-known limitations to those systems which are the stated reasons for developing systemd in the first place. It's reasonable to disagree about the priorities or the approach which systemd uses but instead pekk instead just pretended that there aren't any unsolved problems.

> You're doing it right here: anyone who disagrees with you is a troll

Here, try filling in the blanks: “[ ] is hard to do with SysV init. I don't like the way systemd does [ ] because [] and prefer the way [] does [] instead"

See the difference?

Just because those are the stated reasons for developing systemd, that doesn't mean that those are the actual reasons, nor does it mean that other init systems don't work (even though they could be improved - into, say, OpenRC).

And that definitely doesn't imply that the poster was a troll. You didn't even address that. That is a personal attack, and is typical of how systemd supporters shout down the opposition.

> Here, try filling in the blanks: “[ ] is hard to do with SysV init. I don't like the way systemd does [ ] because [] and prefer the way [] does [] instead"

> See the difference?

So anyone who criticizes systemd and doesn't fill in those blanks is a troll?

The unresolved problems of SysV don't automatically make the systemd approach right either.

We must remember that the BSDs in particular exist in a reality where you can issue a single source control checkout command and you have the whole tree literally to do anything you want with, most of it even with a license amenable to anything.

Linux isn't quite in this world in such a dramatic way as the BSDs, but it exists very close to it.

So "choice" isn't some abstract concept that comes later as an afterthought, it's kind of the whole point in a very concrete and literal sense, in a way that other systems do not offer. Naturally you will see people with their workflows based on that, or some indirect form of that. And these people have working systems, no matter how much people wax philosophical and longingly on forums for "missing" features that are irrelevant to people with already working systems.

So from a certain perspective it's a bit weird to see something like a slide here say that such a system needs to address 99% of people. Who says? It's working fine in its niche.

I think there is also a very simple difference of opinion and difference in personality types here, in a way that I don't often see people acknowledge. On one side you have people who crave authority, order, centralization. One component to manage boot and hardware. One UI toolkit. Choice is bad, freedom is slavery. (OK, just joking on that last one.) Then there are people who just happen to like it the other way. I'm glad that some systems don't bring up a GUI until I ask for one, and even then, it can be whatever I say it is. I find it pleasingly simple that the OpenBSD upgrade instructions basically tell you to run tar and patch. These things are an anathema to those whose personality types are more aligned with, say, OS X, but they work well for me.

As an additional point: in the spectrum between centralization and decentralization, there's a clear difference between Linux and FreeBSD. This is what I highlighted in my post: FreeBSD's greater degree of centralization seems to me to be more efficient and leading to a better integration. That much is a correct observation.

But I did not want to imply that "decentralization equals less choice". As an example: Android is as centralized as you can get for an open source project. That does not stop people from customizing it to bits, so much that it's barely recognizable as Android anymore. The beauty however is that people can do that without influencing the direction of the original project (although that happens to a limited and uncredited degree), or bitching and moaning about it.

That's a fair point. When I wrote "centralization" I mainly meant components rather than development model. Systemd is very "centralized" in that a single package does a lot, but Linux as a whole is very decentralized in development model. BSDs are "centralized" in the sense that they have a single tree in source control, but the main result is a rather small, very configurable base system. (Then you deal with ports and packages on top, which feels more like a Linux distro: scripts that pull distfiles from lots of different sites and apply patches; not unlike the experience of building a .deb from source.)

> It's a shame FreeBSD doesn't maintain an integrated or official graphical interface.

Coming from an OpenBSD background (hey - it's the only OS in the world allowing you run Xorg as an unprivileged user [1]) -- what I'd like to see is an intra-BSD desktop environment based on Google's Material Design guidelines:

http://google.com/design/spec/material-design/introduction.h...

And for ol' times sake, the IRIX® Interactive Desktop User Interface Guidelines:

http://menehune.opt.wfu.edu/Kokua/Irix_6.5.21_doc_cd/usr/sha...

Similar to Quantum OS (Linux):

https://news.ycombinator.com/item?id=8645504

--

[1] http://undeadly.org/cgi?action=article&sid=20140223112426

> hey - it's the only OS in the world allowing you run Xorg as an unprivileged user

Wait, what?

https://imgur.com/ZrdgN1H

This isn't some custom repo or anything, I'm just running stock Arch and have startx in my bash profile. It runs in user mode, and has for about the last half year.

I could be wrong, but I was under the impression that that wasn't the case?

https://wiki.archlinux.org/index.php/Systemd/User#Xorg_as_a_...

  Unfortunately, to be able to run xorg in unprivileged mode, it needs to run inside a session. So, right now the handicap of running xorg as user service is that it must be run with root privileges [...snip...] It seems likely that xorg could be modified [...snip...], and then it could run unprivileged from a user service outside a session.

Thats in the context of using xorg as a systemd user service. You can run startx as a normal user and have the server run unprivileged just fine.