The article points this out. But it also points out (using the Ubuntu release notes) that there are quite a few regressions as well.
In other words, there is a lack of professionalism and polish associated with Ubuntu releases. Maybe they're code-freezing too late in each cycle to be able to stabilize everything in time for the release?
I don't think its about when you code-freeze. I think its more because there are way too many bugs and variations of bugs and not enough people to try and solve them. Which is why all the developers ask for help in triaging and polishing bug reports.
This is especially true when its bugs related to different hardware configurations such as multi-screens with specific graphics cards and so on.
> In other words, there is a lack of professionalism and polish associated with Ubuntu releases.
I agree. I think they should make their releases less frequent, e.g. move to a 2 year relase cycle. I have zero interest in fiddling with a working computer setup every six months in order to get the latest goodies.
Their mission is to get more people using Ubuntu - a 6 month release cycle works wonders for that. People (like me) who think "I could almost use the permanently" get excited, and give it another go, when a new version comes out. Also, it generates a lot of buzz.
However, 9.10 hasn't been a smooth ride for me. When I plug in an external monitor, and open Display Options to enable it, my X server crashes, due to Compiz not handling such a large resolution.
Yes! And this would also help to have a better marketing. News media won't talk every six months about a new Ubuntu OS, but they might do it once every two years. And they would talk about a stable realse. So, we would have happier users and better marketing.
They claim that a poll said only 10 percent of people had a flawless install but the link actually shows 16 percent had a flawless install with another 12.5 percent having only minor problems. I don't know whether those numbers are good or bad. There's a selection bias in the poll (people with problems are far more likely to be cruising the site). I'd like to see a random sample. For myself, I installed Karmic on a quad core desktop and a netbook with no problems at all. The first time I've installed any Unix without any grief.
It's interesting to compare the results of the polls after they stabilise. I'm not sure if it fully accounts for selection bias, but it seems to be a better basis for comparison.
In terms of show stopping problems, Karmic is currently about the same as Jaunty with about 32% of people reporting they couldn't work around their issues, while for Intrepid it was 45%.
I can imagine there's selection bias at that level as well - the people who can't work around problems for one release aren't going to be involved in the poll for the next release.
Anecdotally, everyone that I've convinced to try Ubuntu has been badly burnt by upgrade issues - only my girlfriend has continued to use it since I do the upgrades for her. I'm sticking with it as I seem to be able to workaround or fix anything that comes up, although it feels like I'm rolling dice everytime.
About an hour ago I downloaded the windows installer for Ubuntu. It ran just like any other installer on windows and when I restarted there it was, right beside windows. I can also uninstall it with the add/remove programs window.
That was by far the most painless install of any OS I've ever installed.
Its not just the boot process, but disk access in general will be slower, especially if your disk is fragmented. The file system is implemented as a single file on the windows partition.
I use Debian in a Windows VMware host (at work), and it is more than fast enough. The only problem is that VMWare's X server is broken in such a way as to not be able to draw window decorations if they are drawn via X's standard function to draw window decorations. Very stupid.
I had that problem in Ubuntu and switched to VirtualBox. Surprisingly, when I enabled 3D rendering in VirtualBox and enabled compositing in OpenBox, My performance and snappiness increased.
Packages that are in Debian stable are extensively pre-tested by the users of unstable and testing. I have been an unstable user for years, and the one or two time a year things break (I update approximately weekly) and I try to report a bug, the bug has already been reported and resolved. This is what they call unstable, so imagine how well stable works.
Debian unstable doesn't have releases, so packages that make major changes have to apply cleanly to a variety of configurations (not just whatever the previous version was). Since people are testing these packages daily, any problems are resolved way before anything is marked "stable". These changes are also incremental, so instead of a testing team having to work out 100s of major changes with a six-month deadline, they are made and tested as-needed.
The end result is a system that is very up-to-date but rarely breaks. (Debian stable probably "never" breaks, but it is not exactly up-to-date either.)
I am not sure why people expect reliability from a bunch of unrelated software packages written in C without any automatic testing. Of course stuff is going to break. My laptop's sound card alternately breaks every other major kernel release. Nobody to test + no automatic tests + minor changes in the sound subsystem = breakage. (I don't care about this, though, as the card is junk and I have a decent USB DAC connected to my machine. But still...)
I am mildly surprised every time I see a Linux system boot to completion. (And yes, Linux is my favorite OS.)
Wow, what a useless article. "Early" adopters experiencing bugs?! Clearly the author must be new to "early" adopting. (Why he uses quotes for "early" I don't really understand.)
Because "early adopting" is generally used when talking about beta software, but in this case the software was declared released, but the users' experiences resembled that of early adopters'.
Plus: there are no reports of Ubuntu installs like iPods ;-)
It's also easier to patch software, although it's harder to make it run on hardware someone else designed and who will not tell you anything about how it works. Apple doesn't have this problem either: the least they can do is OSX folks pointing fingers to Macintosh hardware designers.
But I have to admit Apple hardware has the good looks on its side.
The article is a little bit too harsh, but I get the idea. Let me explain ...
First, the title is wrong. Ubuntu is absolutely making Linux MUCH friendlier to use than ever before. In fact I would seriously doubt if Linux could be as popular as it is now without Ubuntu.
But the problems mentioned in the article are real --- every Ubuntu upgrade fixes some issues and breaks some others, especially drivers. I think the reasons are pretty obvious here: Apple has only a small portfolio of hardware configuration and generally Mac users don't tweak their systems as much as Linux folks do, and Apple can actually test each release on all possible hardware (Hackintosh excluded) AND reasonable system variants before they make it public and feel pretty confident it won't break too much stuff. So it's not very fair to compare Ubuntu with OS X for driver compatibility, as there is simply too much combinations of hardware and software to support. In this regard, Ubuntu is more similar to Windows.
But Windows has an advantage right now: it has a much larger install base which forces hardware/software vendors to "manufacture to Windows". And Windows has only a handful clearly-defined "reference implementations" for vendors to test out---currently the "reference implementations" are XP and Vista/Win7, then double the number to account for x86/x64 variants. Even then it is very hard for Microsoft to keep all devices working for all Windows users.
The situation is much worse for Ubuntu: officially there are four editions (U, K, X, and Netbook Remix), and think about the number of platforms it supports. In addition only a few vendors actually test their products on any flavor of Linux. And those who do test on Linux are usually big names like Intel (even then they suck from time to time). Your cheap webcam manufacturers don't really have the money or time or any incentive to test the $5 crap on Linux they sold to you. So the reality is, the test coverage for a particular installation of Ubuntu on a particular hardware configuration for the vast majority of users is pretty limited. You shouldn't really expect too much more than what we have now.
The more I think about it, the more I appreciate the effort Ubuntu and the whole Linux community spent to make Linux more approachable to common folks like my dad (he's been using Ubuntu for quite some time now and never bothered by virus/trojan any more -- I just don't upgrade the system as long as it works).
But problems are still problems. A year ago when my old laptop finally stopped working after a glass of juice got into its body I switched to a metal MacBook. I finally decided that my time wasted on figuring out how to make my webcam work again after each Ubuntu upgrades, if accumulated, is actually worth something. I feel much happier now on a Mac. Sometimes I still long for the power of control on Linux and really get annoyed by various little things on OS X (like even in Snow Leopard the default Terminal.app doesn't support 256 colors -- WTF!!??), but I'm willing to accept the trade-offs. Besides, when I really need the power of Linux I immediately summon the Ubuntu installed in VirtualBox and ssh into it (actually I think this is the preferred way of running Linux -- you avoid all the driver issues by virtualizing the hardware).
I am using Ubuntu 9.10 as my primary OS also and it has been nothing sort of satisfactory. However the article does point out a few useful suggestions in order for Ubuntu to be the leader in Linux Desktop Computing.
I installed 9.10 on my MacBook Pro yesterday. My main issue was that I couldn't get the non-free wireless drivers started without uninstalling and installing a package followed by restarting.
