Hacker News Reader

Comment by acabal | Hacker News Reader

Nope. Ubuntu won't be ready until upgrading Firefox (or any individual package) doesn't force insanity like opposite window controls or Unity without asking. Regular people aren't interested in seeing drastic and seemingly nonsensical changes like that for no reason, and almost certainly aren't interested in being part of some grand OS experiment. They want something that is consistent and works.

While package management is an excellent concept, the way Linux packages are handled--with shared dependencies--is what engenders the ridiculous milestone-distro system that I think is keeping desktop Linux back. Having to swallow upgrades to every single package on the system just to get one package (like FF) upgraded means that instability and unwanted change--exactly what regular users hate--will be ingrained in the experience.

The solution, I think, is either to get rid of the shared-dependency concept and move to a more Windows-like static install system, or form a truly stable rolling-release distro. The only big one is Debian/Mint, but from what I hear that's still too unstable for even power users.

View on HN

This is one of the major reasons I don't suggest Linux to anyone. There was a time I'd get people to switch from Windows to Linux; now I tell them to either upgrade to Windows 7 (if they're not on Windows 7 already) or get a Mac.

I'm also concerned about three secondary issues. One, there is a lack of strong leadership in the Linux community. Shuttleworth is the only person who can shake things up, and whenever he does everyone whines as if Shuttleworth had caused them some grave personal injustice.

Second, of course, is fragmentation and the fact that Linux developers change things needlessly (see GNOME 3, KDE4, PulseAudio, etc.) The end result is different distros doing things in slightly incompatible ways. These days I see more and more people package their software for Ubuntu only, and provide a generic .tag.gz for other distros. I'm no Ubuntu fan, but I jump for joy every time I see this. Distros need to die.

Third is politics. I know people who contribute to a few high-profile open source projects and I've watched them waste their time on bikeshed discussions on mailing lists and IRC. I've even watched one guy work on a project for three months only to have the design team discard it because they had a different idea in mind. The result was a long battle between programmers and designers. I guess this is a direct result of lack of leadership.

I've been hearing talks about a GNOME OS. I sincerely hope the GNOME folks go ahead and build it. As I've said before, distros need to die. So much manpower is wasted on just packaging things for different distros. From a consumer standpoint, choice is okay if you have to choose between two or three things, each with their clear merits and demerits. Choice is fatal if you have to choose between ten things which are all nearly identical.

Sometimes I feel Linus is the only guy who knows how to run a project properly. He knows how to say "no". If only we could coax him into building a distro ...

"There was a time I'd get people to switch from Windows to Linux; now I tell them to either upgrade to Windows 7 (if they're not on Windows 7 already) or get a Mac."

What's changed?

"Shuttleworth is the only person who can shake things up, and whenever he does everyone whines as if Shuttleworth had caused them some grave personal injustice."

People get comfy in their distributions. When Shuttleworth dictates "design" decisions with the reasoning of "just because," and as a result many people no longer find their distribution of choice comfortable, can you really blame them for being pissed off?

I know many people in this situation since Ubuntu 11.04. I'm one of them. I wasn't happy with Unity. I had to spend the time to migrate my desktop and laptop to Fedora, and get readjusted. This was annoying, but at least I'm comfortable again.

"Second, of course, is fragmentation and the fact that Linux developers change things needlessly (see GNOME 3, KDE4, PulseAudio, etc.)"

Fragmentation is inevitable in free software. People build what they want - you can't stop them, and you don't have to use what they make. The solution is not to provide the "one distribution to bind them all" and somehow dictate that everyone should use it, but to get the different ecosystems and environments following standards. They're doing that pretty well.

Anyway - you wish for "strong leadership" in the community. There will always be more than one center of strong leadership (because of personal preferences), and hence, fragmentation results.

"These days I see more and more people package their software for Ubuntu only, and provide a generic .tag.gz for other distros. I'm no Ubuntu fan, but I jump for joy every time I see this. Distros need to die."

It seems you wish for the same universality for Ubuntu as exists for Windows and Mac OS X. Do you know why those platforms are universal? Because it's relatively difficult for graphical programs to be ported between them. That problem doesn't exist for the free software ecosystem thanks to portable graphical toolkits, freedesktop.org, etc. Any good software written for distro X will soon be running on distro Y (if it doesn't already from the get-go), and that will never change.

"Third is politics. I know people who contribute to a few high-profile open source projects and I've watched them waste their time on bikeshed discussions on mailing lists and IRC. I've even watched one guy work on a project for three months only to have the design team discard it because they had a different idea in mind. The result was a long battle between programmers and designers. I guess this is a direct result of lack of leadership."

I agree that this is a huge problem. Design by committee is inefficient and ineffective. I think it would be a good idea for GNOME to have leadership elections at the end of each release cycle, to get someone to lead the GNOME project through to the next release.

"I've been hearing talks about a GNOME OS. I sincerely hope the GNOME folks go ahead and build it. As I've said before, distros need to die. So much manpower is wasted on just packaging things for different distros."

By those distro maintainers' own initiatives. They do it because they want to.

"Sometimes I feel Linus is the only guy who knows how to run a project properly. He knows how to say "no". If only we could coax him into building a distro ..."

Linus is (a) busy, and (b) has no clue about usability. He runs Xfce, for heavens' sake :)

This is such a great reply. Thank you, majika.

> What's changed?

Nothing much on the Linux side; Windows has just become better. Windows XP as I remember it was utter shit. Windows 7 is actually pleasant to use. I also got myself a MacBook in 2008, wrote many Cocoa applications, and found that the Mac was a better work machine than my Linux machine. So if someone can't afford a Mac, I tell them to use Windows 7. Otherwise it's time to hit the Apple store :)

> People get comfy in their distributions. When Shuttleworth dictates "design" decisions with the reasoning of "just because," and as a result many people no longer find their distribution of choice comfortable, can you really blame them for being pissed off?

Totally agreed. So can we say that not only does the Linux world lack leadership, the little leadership it does have makes terrible decisions? :p

> Fragmentation is inevitable in free software. People build what they want - you can't stop them, and you don't have to use what they make. The solution is not to provide the "one distribution to bind them all" and somehow dictate that everyone should use it, but to get the different ecosystems and environments following standards. They're doing that pretty well.

Once again you make a very good point. But I'm looking at this issue from a consumer's standpoint. The consumer wants to deal with one entity; he doesn't like fragmentation. In any case, these mostly identical distributions serve no purpose. Most of them have no reason to exist. Personally, when I want to use Linux -- and I do use Linux on servers because that is where it shines -- I default to Debian and pretend that nothing else exists. I'm not good at dealing with choice, and I believe the man on the street is almost as bad as I am.

> It seems you wish for the same universality for Ubuntu as exists for Windows and Mac OS X.

I do. I honestly do. Universality is a good thing. I know a properly written GTK+ app will run identically on every Linux machine out there, but there are still some very tiny incompatibilities, especially w.r.t. packaging and distribution. As an indie developer, I don't want to deal with that stuff. I just want to create a DMG/MSI, ship it and expect it to work for the next 5 years without issues. The universality of Windows and the Mac gurantees that.

Oh, one more thing: Windows and OS X let even bad developers bang out working applications. All their assumptions about their OS, their hardcoded paths, their assumptions about certain libraries existing at certain versions, all their bad code ... it all just works. Small point, but I think it's an important one. All of us were ignorant once :)

I know I'm going to sound like a crazy person here, but I just feel there is something very wrong with fragmentation. I don't know; it's just something I feel. I guess I probably am a crazy person ...

> Linus is (a) busy, and (b) has no clue about usability. He runs Xfce, for heavens' sake :)

Hey, Xfce is pretty damn good :)

I would advise the ordinary non-technical user to go Mac. Fewer headaches all around, largely through restricted choice. The platform drives me bananas. Windows makes me postal though.

Linux / Free Software has strong leadership, but that leadership generally has limited scope. The GNOME team have certainly taken bull-headedness to new heights within their own limited realm of control. To somewhat muted applause as I recall.

OpenBSD (granted, not Linux) isn't particularly renowned for having a shrinking violet at its head. Nor Linux itself, nor the FSF. Other major projects have a mix of leadership styles, itself part of the (r)evolutionary culture of trying different technical, social, organizational, and leadership structures.

That diversity (including various ill-advised developmental branches) is part of Linux / Free Software's strength: a great many parallel paths are being tried. Some will succeed, some will fail. In general, the better alternatives generally at least get a good, if not fully fair, hearing.

Politics is inevitable when you're dealing with more than one individual or organization. Grow up and get over it. You'll note that these same political discusses exist in the proprietary world, you're just not (usually) party to them.

The proper counterproposal to a bikeshed argument is to demand construction of a nuclear power plant.

Those "creative differences" are called "forks" and again, are a key strength to Free Software (see previous comments).

> Having to swallow upgrades to every single package on the system just to get one package ... [means] ...instability...

Just curious, have you really experienced instability by installing all updates on any mainstream distro? If so, could you share what happened? I've not had that experience but would like to know about it.

Yes. For example when Ubuntu switched to Pulseaudio, it borked my sound. Even with the release after that one (can't remember which one it was), I had to have a shortcut on my desktop for 'killall pulseaudio' because just playing music for more than 15 minutes would end up in garbled sound. Or in Karmic, when a regression in the Intel video drivers made even dragging windows stutter on my laptop with 4GB of RAM. Or when Lucid broke hibernate for my laptop. It was fixed in Maverick, then broke again in Natty. Literally every distro upgrade of Ubuntu for me has resulted in more than one serious regression in something that used to work.

Ubuntu and Fedora, though the most common desktop distros, also are notoriously NOT very stable. Fedora is RedHat beta-test playground, and Ubuntu is built from a Debian-unstable snapshot. OTOH many people complains that Debian, RedHat or CentOS are too long to implement the latest features, but this is the price to pay for extreme reliability. Know your needs, choose accordingly.

I just wish there was a middle ground though where you dont have to choose between extreme reliability + years of waiting and fast turnaround + notorious instability

Debian testing is both reliable enough (I think there were no more than 2 cases when an upgrade would lead to an unusable state in the past 8 years) and not too far behind the bleeding edge. Or you could use a rolling release distro such as gentoo, arch, or in a related world FreeBSD.

The problem with Debian Testing is that once something breaks badly, you usually have to wait two weeks until it is fixed, because changes need to trickle down from Sid. In my experience, Sid is more useful as a desktop. But it's certainly not for the beginner.

For desktops: Debian testing + pinning as needed to backports, sid, and experimental.

For servers: Debian stable + backports.

That last in particular is vastly preferable on servers to the usual "Oh, we've got to run Red Hat because it's industry standard, but we're going to have to rip out the guts and install a metric shit-ton of third-part / roll-your own / nonstandard packages, and then hope that the poor third-generation admin to inherit this cesspit can make heads or tails of it.

Since my favorite Debian combo was already mentioned, I should mention my usual plan-B: CentOS with CentOS Plus enabled on the server. What flavor of Unix you run on the desktop is irrelevant, although I frequently find myself in dependency hell when I am working on a Mac. Installing the qtconsole version of iPython is next to impossible.

It has happened to me once in a while. In fact my full upgrade to (k)ubuntu 11.04 ended up leaving the machine in a unbootable state. Not a big deal since I had data backups. But still time consuming.

+1 here. mysql-server is compltely broken in 10.04LTS.

First: Linux does expect (and reward) a certain level of technical literacy. If you don't want to acquire or exercise this, well, you have alternatives and you've mentioned several. Those do come with trade-offs of their own.

If you don't like system upgrades for single-package updates, learn to use your distro's 'backports' feature. What? Your distro doesn't have backports? Shame.

Or switch to Gentoo and install/upgrade just the packages you want to use. Yes, you'll have to learn how to compile shit.

The UI in Linux is highly modular. You can switch to a different (or no) desktop environment pretty easily. There's a switcher in your log-in manager. Try it.

There's a reason for package dependencies, and there's a huge payoff (single-point-of-control whole-system upgrades with very low failure risk). It's not perfect, but on systems which have developed a mature packaging system over time (Debian, derivatives), it works exceptionally well, and provides access to over 30,000 packages.

Firefox is upgraded as soon as it's released by Mozilla in all supported versions of Ubuntu (except 10.04, which will be upgraded when Mozilla EOLs 3.6.x for good)

Even with 10.04 you can use the FF PPA for the latest version.

> The only big one is Debian/Mint, but from what I hear that's still too unstable for even power users.

You're forgetting about Arch, which is absolutely amazing. Simple, yet powerful.

I'll second that. Although you can get a similar simplicity in many other distros, however, what really makes Arch great for me is the community; they have one of the best wikis I have ever seen.

The thing I hate about Arch: the AUR. It's horrible. There is no official tool to handle it, the package stability or coverage is spotty at best, and the unofficial tools (yaourt, etc.) suck.

But you have to use AUR, because that's where the breadth of the packages are. Arch's team is relatively small (compared to Debian's anyway) and they can't maintain that many packages. I agree with their choice of letting the community handle the rest — from their perspective it makes sense.

From my perspective, I'm never returning to the nightmare of AUR again. a) if I want to wait for stuff to compile for ages and then fail, I can just use Gentoo. b) Gentoo is at least more or less stable and has good coverage/stability. c) Portage is at least a good package manager (the second-best in Linux-land, imho, with dpkg/apt-get leading the pack.) Not to say pacman is bad, but yaourt just isn't good.

Can you give some examples of must-have software that are only in AUR? Having use Arch for years on multiple machines, I've rarely had to use AUR. And when I do, I've found it to be a marvelously simple mechanism - a bit like homebrew on a Mac, where you can easily edit the recipe if needed. It sure as hell beats installing unsupported software on Ubuntu.

Debian Stable with Backports[1] seems a decent solution to me.

[1]: http://backports-master.debian.org/

I'm guessing that you also think web apps are also not ready for the average user? After all, there you don't even get to have a say when the upgrade happens. And you have no way of delaying, cancelling, or undoing it.

As far as users are concerned, random stuff changing for reasons unrelated to their actions are all the same.

Upgrading Basecamp doesn't force an upgrade to Gmail. Your comparison doesn't really apply here.

Does that matter? In both cases, stuff still gets changed under you at more or less random times. With web apps, you don't even have control over when random stuff could possibly change.

In Linux terms, that would be like your package manager constantly and silently updating everything in the background all the time. How dependent or independent packages are doesn't matter in this sort of scenario.

I'm not entirely sure that it makes much of a difference. Chrome shows that gradual silent updating works for local apps. Linux needs to handle that better, sure, but I think that's the way to go.

Chrome's updates have been nothing like KDE4 or Unity or Gnome3.

Also, people have also revolted against and deserted versions of Web Apps like Digg v4, so it's the experience that matters, not just change in general. There should be no need to foist updated versions of other apps just to get a version of another app.

I haven't seen an app that depended on KDE4, Unity, or Gnome3, or couldn't be run outside of them.

Gradual silent upgrading is poised to be a huge problem for users and yet another reason I stay away from Chrome. You should never impose an upgrade on a user, at most just strongly suggest they install and make it easy if they choose to do so.

Upgrades have the potential to break things users may find critical. Say for example in a browser, User A uses a plug-in that they find essential to their daily work and it hasn't been updated to the latest release (developer is on a sabbatical roaming the world for 2 months). Now you force an upgrade to their machine and you break their plug-in leaving them with no recourse. This is only one slim example but it's potentially a huge issue. A couple years back a Windows auto-update once killed a client's server for 2 days until I could identify the culprit. There at least I had recourse to uninstall and reject the update, with something like Chrome I don't.

In a web app things don't inter-relate the same as they do on a local OS so it's mostly a non-issue outside of user experience. Forcing upgrades to locally installed software is just asking for trouble all around.

> Nope. Ubuntu won't be ready until upgrading Firefox (or any individual package) doesn't force insanity like opposite window controls or Unity without asking.

Last time I checked, Ubuntu kept existing setup, desktop etc. when upgrading. Anyway your rant is nonsensical: where is my good ol' "Start" menu organisation when I upgrade to Win 7?

> The only big one is Debian/Mint, but from what I hear that's still too unstable for even power users.

I'm speechless, you seem to know nothing of Linux but quick to make judgements. Debian is the most stable (too stable, may be argued) system; I have several machines that were upgraded flawlessly from 3.0 to 6.0 without ever breaking. On the other hand, if you like rolling release, there are many distros that work just like this, like gentoo, archlinux, etc.

> Last time I checked, Ubuntu kept existing setup, desktop etc. when upgrading.

Have you ever dist-upgraded an Ubuntu distro? Upgrading to Lucid moved your window controls. Upgrading to Natty turned on Unity. While it's true that upgrading from, say, XP to Win7 will change your start menu, the point is that you can still upgrade your "packages" on XP, WITHOUT having to get the new Win7 start menu.

> I'm speechless, obviously you're babbling without even knowing what you're talking about; ignorance and hate won't lead you anywhere.

Obviously YOU don't know what you're talking about; Debian Mint (LMDE) is based on Debian-testing, which is supposedly HIGHLY unstable.

You're doing a dist-upgrade, which is exactly like going from XP to Win7. Stay on LTS if it bothers you that much.

That's precisely my point. In the Linux world, your distro version is inextricably linked to the versions of every major and minor software package in a way that just doesn't happen in Windows.

If I stuck with Lucid, I'd still be using Firefox 3.6. (Yes you could add a PPA, but opening the terminal is beyond the technical skill of the average user.) Linux won't succeed on the desktop until the phrases "Upgrade Firefox" and "Potentially bork my entire system with unexpected regressions" don't go hand-in-hand.

FF will still auto-upgrade to 6 even on an XP machine.

> your distro version is inextricably linked to the versions of every major and minor software package in a way that just doesn't happen in Windows.

What's the big deal with versions? It's not like it's too much trouble updating Ubuntu from one version to the next (you know you can turn off Unity). Most of the time, you can even continue working while it upgrades. This is not like Windows where you have to reboot the machine with the install media and wait a couple hours to get the new version.

And while the Firefox teams does an amazing job making FF 6 for XP, the same can't be said about the Internet Explorer team. If you want the latest IE, you need to upgrade Windows.

Ubuntu has a security policy to upgrade supported versions to the latest supported browser version. So, in Lucid, if 3.6 is still supported by Mozilla, that is the version still on maintained (you know, for consistency...thing thing most people here are railing against).

However, as soon as mozilla drops support for 3.6, the latest supported version is installed. This happened for Lucid, as I understand it.

This is a sane policy, IMO.

If, however, you are on Lucid and want to run a different version of Firefox, it is rather trivial. You can find it in Ubuntu Software Center after adding a repo and install it. This is NOT what most non-technical people want to do, however, and that is why the policy makes sense IMO.

And, I agree, FF auto-upgrading itself to 6 is nice on OSX and XP, but that is a Mozilla thing, not Linux. Mozilla makes that difficult to do and the various linux distros need to deal with it.

> the point is that you can still upgrade your "packages" on XP

Like DirectX 10 or 11? The fact that you can upgrade is due to the abnormal shelf life of this particular release of windows. Try upgrading most programs on a win98SE or win2K machine.

> Obviously YOU don't know what you're talking about; Debian Mint (LMDE) is based on Debian-testing, which is supposedly HIGHLY unstable.

Mint is based upon Ubuntu, which is based upon debian-unstable (Sid). Debian testing hardly is highly unstable; most debian users run testing on their desktop systems, and keep stable to the servers.

I think you're misunderstanding what I'm referring to. I'm specifically referring to rolling-release distros, of which LMDE is one. It's based on Debian-testing.

Regular Mint is indeed a milestone-based distro built on Ubuntu. But LMDE is a rolling-release based on Debian-testing, with no relation to Ubuntu. That's specifically what I'm referring to.

There are certainly other stable rolling-releases out there, but I think everyone can agree that a distro like Arch is about a lightyear away from being Grandma-friendly (which Ubuntu more or less aims to be).

> It's based on Debian-testing, which is more unstable than Debian-unstable

Wrong. Testing is more stable than Unstable, which is more stable than Experimental. Furthermore, Ubuntu does extremely heavy modification to the Debian base, repackaging quite a number of it's core packages, and replacing a number of central parts at times (eg, the boot system).

If you're going to say that people don't know what they're talking about, please make sure you do.

You're right, I've corrected my post. But I still stand by my general point :)

Reply on news.ycombinator.com