Although I started out using SUSE (now OpenSUSE), I’ve been using mostly Ubuntu since 2006 and no other distribution proved as reliable as that one, particularly starting with the April 2007 (Feisty Fawn) release. I found that it detected my wireless card and downloaded the correct driver for it and that the desktop was conservative and didn’t get in the way. GNOME 2 had been in development since 2002, and although the very first version was lacklustre (I remember well the slow screen-redrawings of that version, whenever you moved or closed a window), it had developed into a very solid and reliable desktop. Along the way, KDE radically re-invented itself and after KDE 4 was first released, it took a couple of years for it to stabilise and become reliable, losing a large number of followers along the way when they did not recognise, much less like, what they saw. In addition, most well-used apps, many of them also available on Windows and the Mac (such as Firefox, Thunderbird and, more recently, Chrome) used the same application toolkit (which provides windows, buttons etc) as GNOME itself (called GTK+), so it looked well-integrated. Of the current generation of Linux distribution, only Ubuntu manages to replicate this (by updating its theme for the latest version of GTK).

Come the end of 2010, however, the GNOME developers had decided to radically re-work their desktop, and what they came up with was a “shell” in which everything except the application windows and one top panel are hidden, so as to maximise screen space for applications. At the same time, Ubuntu rolled out Unity, a Mac-like set-up with a menu across the top and a “dock” (or launcher, as they call it — not that the term “dock” is trademarked, as it has been used in open-source projects since the specification for OpenSTEP was open-sourced years ago) down the side. These had the advantage of maximising vertical screen space on small, wide screens such as are common on laptops, but the latter, in particular, does not suit screens with a high vertical resolution (such as my desktop screen). Although the launcher can be retracted, it cannot be repositioned or resized. As part of the design project (called Ayatana) it was part of, the system tray was crippled as its use was determined to be inconsistent and (supposedly) confusing, so apps that depended on it (like the one I develop, along with most KDE apps) were prevented from functioning. And other bugs started appearing, such as a failure to detect my wireless card, which had worked fine up until the April or October 2010 releases. It didn’t in either of the 2011 Ubuntu releases.

GNOME 3 featured the already-mentioned Shell, as well as a fallback mode which slightly resembles the old GNOME, if done right (which it often isn’t as it’s an afterthought). The Shell hides everything except a top panel featuring a clock and a calendar; you use the Windows key (or move the mouse to the corner) to activate the “Activities” screen, which allows you to start any application, choose another virtual desktop (similar to the Mac’s Spaces) or use its Dock (yes, it has one now as well) to launch a commonly used app. The only problem is configurability is poor — the basic product does not even come with a configuration tool to change the fonts used — and it does not offer an easy way of turning off annoying features such as “edge tiling” (a window shrinking to half the size of the screen when moved to the edge — it seems not to appreciate that you just might have wanted the window to be at the edge). Although it offers a system of extensions (which do things as simple as allowing you to shut down the system from a desktop menu, which you could do in the old version of GNOME), many of these (including the simple one just mentioned) will actually cause the desktop to crash. At least one “reworking” of GNOME Shell is heavily based on such extensions — I installed Linux Mint 12, after finding that the latest version of Ubuntu would not even play sound on my laptop — and I found that, although it bizarre screen effects when switching applications (in particular, when switching to Qt apps, such as mine). Regarding the lack of configurability, an editorial response to a letter in Linux User and Developer magazine mentioned a “developer knows best” attitude, in which desktop capabilities are limited by the tastes and design decisions of the developers.

Currently, on my laptop, I’m running KDE on Linux Mint 12 (as Mint is based on Ubuntu, you can download software from Ubuntu’s repositories, and from any third-party Launchpad repository also). KDE is on the face of it, nowadays, a very stable and presentable desktop, which has regained the usability it lost in the big move to KDE 4. I hardly ever use its two special features, Activities and Plasmoids, except for the folder displays on the desktop background (a good alternative to the automatic display of the contents of the Desktop folder on the background). However, I generally find that I do not use any KDE applications when I’m using KDE, as many of them are not up to scratch (there are exceptions, such as the photo manager Digikam and graphic editor Krita). The mail client (KMail, which can also be used as part of Kontact, which also offers a calendar and various other organiser-type features) is especially awful: it continually threw up errors when trying to configure a simple IMAP email account (giving the impression of failure, and indeed it kept exiting), and continues to do so during use. I use it, in fact, only because Thunderbird refuses to display the contents of some messages (which I am certain is a problem in Mint, not Thunderbird, as I have never come across this problem with Thunderbird before and it is not a problem with OpenSUSE or, for that matter, Ubuntu).

In short, a year ago, Linux (and mostly Ubuntu at that — I tried others, but always came back to it) was a reliable operating system, fairly easy to use if you had a bit of computer experience, with a well-maintained desktop on which nothing looked out of place. Today, we have a hotch-potch of ill-conceived desktops with important features missing and some unwanted, forced on us by developers who believe they know best and did not think to consult with application developers outside their circle, and the people behind the biggest distribution even seems to be neglecting hardware compatibility. The reason, some have alleged, is the drive towards making the desktops compatible with the small-screen netbooks, yet this is no reason to neglect the traditional desktop user or screen formats other than cinema aspect (it may also be that, with the rise to prominence of the Linux-based Android on mobile phones and tablets, securing a place for Linux on the desktop has become less important to many people). The upshot is that Linux is no longer the pleasure to use that it was a year and a half ago (while at least one of the commercial rivals has finally put out a usable and reliable operating system) and my use of Linux is likely to drop considerably as of next week.

Until last year, Ubuntu’s main desktop was GNOME, which had been around since 2002 and had developed over the years into something very stable and reliable. That wasn’t the main reason it succeeded in that; the main reason for that was that its competition in the free software desktop world, KDE, was based on a toolkit called Qt which could be used for free and open-source products but not commercial ones without a paid licence. The toolkit used for GNOME (GTK) could be freely used for either, although I do not recall many heavy commercial applications being ported to Linux using GTK (RealPlayer and VMWare used it, though). While Qt was released under similar terms to GTK after Nokia took over the company that developed it, KDE’s developers then radically re-invented their product in 2005, causing it to lose a lot of ground for years afterwards. By last year, however, KDE had returned to form and GNOME had come to be seen as a bit long in the tooth.

Both products offered essentially the same things — menus and applets which could be arranged on one or more panels, and these were often arranged so that they resembled a Windows desktop; GNOME more often resembled a cross between a Mac and a Windows environment. Interestingly, perhaps because display technology has improved and more computers have accelerated graphics cards that are capable of more sophisticated 3D effects, the new GNOME environment has moved to a more Mac-like feel. The new version of GNOME has a “Shell”, offering a “Dash” (obviously modelled on the Mac’s Dock), and windows that scale down at the touch of the Windows key so that you can see all of them (similar to Exposé on the Mac), although some of its other improvements are quite original, and not all of them are a great improvement. It still offers multiple desktops, for example, but using them is more involved than in KDE or the older GNOME.

Ubuntu does allow you to install GNOME 3’s Shell, but defaults to its own shell called Unity, which was originally based on the old GNOME and appeared on the Netbook version of Ubuntu from 2009 onwards, but became the main desktop for its main release for the April 2011 (Natty) version. It offers an even more Mac-like user experience, with a top-of-screen application menu and a permanent Dock (called the Launcher) at the left hand side of the screen. The trouble is, it doesn’t go the whole way. You can’t move the dock from the side, for one — which may suit wide-screen monitors fine, and most computers (and nearly all laptops) come with wide screens nowadays, but some of us are still suck on 1280x1024 displays, like me, and a strip down the side of the screen, even though the screen is wider than it is long, takes up what looks like an unreasonable amount of screen space. A strip along the bottom would give more space for icons and allow apps the full screen width, giving something more like the cinema aspect that is increasingly standard. Besides which, it’s a choice, which the Mac offers and Unity does not.

Even more irksome is the fact that you can’t program the launcher as you can on the Mac — you can simply set up a menu and apply it to the Dock icon with one command, and it will appear if you right-click (or long-click) your application’s icon. That doesn’t work on the launcher — you can add menu items to it, but although there may be a way of doing this from within a program (which is not well documented if it exists), but the standard way is to write a file with a set of commands that can be run whether the program is running or not. All very well, but some way of automatically running the program first if it isn’t running already would be very useful (as it is, you have to write a script yourself to do that). It still isn’t as much functionality as you get with a menu you can attach to the system tray icon (which Unity has been busy crippling), and it’s not as simple to program. GNOME 3’s Shell lacks even this functionality at present. Worse, it pops up a menu saying “New window” which does no such thing — it just starts a new copy of the program, which may not actually open a new window at all. The “Quit” item in the same menu in Unity is just as bad. It doesn’t actually close the program, just all the open windows. If the app has a system tray icon, that remains visible, i.e. the app is still running.

Canonical (the company behind Ubuntu and Unity) have been trying to get developers to address what they saw as the mess that was the system tray and GNOME panel structure. To be honest, it was inconsistent, as they say, with some items being buttons and others being menus, there were different stylings, and accessibility problems. However, it does provide a way of providing a global menu for a piece of software and it is a cross-platform standard, so apps written using it with one toolkit can run on all the major desktop environments, plus Windows and Mac OS X. Since Unity’s “app indicators” aren’t present on the Mac or Windows, an app that depends on that alone is tied to Unity — it will not even function properly on any other Linux distribution, or on KDE. Canonical acts as if its product is the Linux desktop, when in fact it is merely one version of it (albeit the single most popular) and Unity does not ship as standard on any other version of Linux. I’d be the first to admit that the Mac’s way of providing a global menu through the Dock icon is better than using the system tray, and if Unity had provided the same, I would have used it some time ago. But it hasn’t.

Canonical have some other funny design concepts, among them doing away with an explicit Quit command. They present the Quit command as a compromise dating from back when the only GUI was on the Apple Lisa, which had a 5MHz processor, 1Mb of memory and two floppy drives and only one application could run at any one time. More recently, they claim:

Mac OS X moved the “Quit” menu item from the “File” menu into the junk drawer that is the application menu. But quitting has remained a basic part of how Mac applications present themselves, and it has persisted for many applications in Windows and Ubuntu as well.

A few behemoth applications, such as LibreOffice and Gimp, still keep “Quit” separate from “Close” for the original reason — to save you from having to wait for the application to relaunch after closing its only document. But that is fixable, and all other applications have become fast enough that they don’t need it any more. After all, they’re running on hardware that is hundreds of times faster than it was in 1984.

How is the Mac application menu a “junk drawer”? Having the Quit command there, rather than the File menu, is a quite logical thing to do if the application menu exists. It doesn’t anywhere except on the Mac, of course, but that is their way of making the Quit command look like a throwback. They diverge into how modern environments such as smartphones do not have the Quit command. I have an Android and I can see the consequences: apps that just won’t die, and annoyingly keep throwing up notifications long after you last used the app. Some Android apps do, in fact, have a Quit command. There are still applications, even on desktops with dual-core processors and 4Gb of RAM, that are inefficient, bloated memory and processor hogs and being able to get rid of them when finished with them is not a throwback, it’s essential.

All that says, Unity works fairly well on my laptop. There are some nice features, including the Ubuntu Software Centre which starts if you try to load an application that is not there, giving you the option to install it; however, it should really have an easy way of running it from USC once you have installed it. On the 64-bit desktop machine, for some reason there are bugs which don’t seem to appear on the 32-bit version on my laptop: Google Chrome always opening to cover the whole screen (maximised), and regaining the GNOME title strip when you click the button that sizes it down (a known bug that has existed for a long time). On the laptop, it covers everything except the dock. That doesn’t happen in Unity 2D, but when you scale Chrome down there, it does not redraw properly and so bits of the old bigger window are still there until you move something over them. The icon for the text editing program GVim is missing on the desktop, but is there on the laptop. On my laptop, although it detected the Broadcom wireless modem and downloaded what it thought was the correct driver for it, the driver did not work and it took a browse of the Ubuntu forums to get the right packages. This bug was present in the last release as well.

On both machines, speed and responsiveness is good. I used the “desktop CD” to install on my laptop; it was incredibly slow and unresponsive, opening up with the installer program that I didn’t want to use just yet, and crashed when I tried to close that. It then started up the normal live-CD session, and when I ran the installer again, it failed while cleaning up after finishing the install, giving no indication of how to fix the problem. I rebooted into the newly installed operating system and nothing was amiss, however. On the desktop, I used the alternate installer CD (a straight text-based installer) and everything worked perfectly. There was, as expected, a huge update to be done afterwards; this is because I installed from a Beta CD, but by that time, the packages had been updated to the Release Candidate versions but no new CD had been released. I have always preferred to use the text-based alternate installer CDs and will continue to do so.

In short, an improvement on last time. I haven’t tried out KDE on this version yet, but might well do as none of the four versions of GNOME are quite satisfactory on the desktop. On the laptop, this one is a keeper.

Fedora 15 was the first official distribution release to actually come with GNOME 3 as standard and not GNOME 2, as all other releases this year have. I downloaded the live CD and proceeded to run it, and found it impressive to begin with. GNOME 2 provided configurable menus and panels, and you could have as many panels as you liked and put the menus, clock, system tray, window switcher and various other gadgets wherever you like. GNOME 3, or at least the GNOME Shell which is currently the only way to run it, streamlines everything, with a single panel at the top containing a clock, a “message tray” containing a volume control, a network monitor, an accessibility control and the user menu (which lets you log out or open the settings box).

Everything else is hidden, and if you move the mouse to the top left-hand corner or click where it says “Activities” in the top-left corner, GNOME will shrink down all your open windows so you can see everything, show you the Mac-like dock, the menu you can use to launch new programs, the bit of the system tray that showed application icons and the virtual desktop switcher (similar to the Mac’s Spaces). As on the Mac, you can add programs to the dock that you use all the time or remove what you don’t. Also like on the Mac, it seems to be possible to add a menu to the dock icon (or to the application icon when it appears in the top panel), but most applications do not support this at the moment. In addition, they do not interact well with applications that are based off the system tray; if you choose the option that says “Exit this app”, it will just close the window, not the program.

The problem is that the new desktop is so limited. GNOME seem to have taken the opposite line from KDE when it launched KDE 4.0, which included desktop applets, whole new frameworks for searching and personal information management and hardware interaction, a brand new theme and a lot of use of transparency. In short, everything was brand new and untested. GNOME seemed to have taken the line of removing features and then adding them back as requested (one hopes). The trouble is, so much has disappeared, including, for example, the desktop preferences window, so you cannot alter the fonts or change the window stylings. The standard user menu also does not let you shut the machine down, which you can do from every other desktop on Linux or any other operating system. There is also no obvious way of setting how many virtual desktops there are, something that was easy in GNOME 2 and remains so in KDE 4.

A further annoyance is that activities mode only lasts until you do one thing, like launch a program. If you want to launch two, you have to go back into activities mode, and again if you want to launch another. With the old panel system, you could drag an application icon from a menu to the panel, so that you could launch it with one click. It would be useful to have some way of keeping activities mode on until you choose to turn it off. Also annoying is the way the system tray disappears into an obscure pop-up at the bottom right; you have to move the mouse to the bottom right to see the icons. It appears that someone, somewhere has decided that system tray icons are a bad thing, but neither GNOME 3 nor Ubuntu have implemented a satisfactory alternative. As a Qt developer, I can easily set the dock icon menu to be the system tray icon menu and not display the latter on the Mac, but there is no obvious way to do this on GNOME 3 (or Unity). Until it’s that easy, making the system tray this obscure is a bad idea.

As it happens, a number of extensions are already available to put back some of the functionality that is missing in GNOME 3. Fedora, however, has played along by removing some of its own graphical configuration tools, like the one that you could use to set the SELinux security level. SELinux (Security Enhanced Linux) is the security monitor which comes with Fedora and can block actions it deems a threat to your system’s security, but any time I have run Fedora with SELinux enabled, it has produced alert after alert for things that no other version of Linux noticed, and I ended up disabling it. There always was a graphical tool to do this; there is none in Fedora 15, which has regularly resulted in the web browser, Chrome, crashing. I ended up changing this by manually editing the SELinux configuration file.

For all this, however, GNOME 3 is quite usable. It’s stylish and what is there works very well. I have not had it crash on me once (the Chrome problem excepted, and that was just that one program). However, it’s stripped very bare and generally, early adopters tend to be adventurous users, not those who want a very basic and simplified desktop experience. If you want to install Fedora 15, it’s either this or KDE 4, and I recommend the latter. Otherwise, it is not worth going to extra hassle (such as by enabling unofficial repositories) to get GNOME 3 now if your distribution offers the latest version of GNOME 2. It’s probably better to wait until some of the old functionality has been restored.

What I’ve installed is a Beta, but the most objectionable features seem to be “by design” rather than bugs. As a long-time Mac user (although I don’t use my Mac much now as it’s out of date, slow and expensive to replace), I’m more than familiar with the whole concept of a Dock, which has icons that can both start and control applications. The problem is that their launcher is pinned to the left hand side of the screen and can’t be transferred to the bottom, which is where it normally is on the Mac. The problem with that is that not everyone has wide, cinema-aspect screens; my screen is 1280x1024, which is still wider than it is high, but having a launcher fixed to the side is much less convenient than having it at the bottom. There seems to be no way of changing this.

It’s also impossible to customise how applications launch: if you prefer a bigger Terminal window than normal, there seems to be no way to change it to what you like, which you can do when launching an application from a normal panel icon. Unity insists on launching large applications maximised (i.e. covering the whole screen), rather than the size it was when you last closed it. This is also apparently a design decision, but it interferes with my right to decide how I use programs on my system. There also appears to be no way of correcting this.

Worst of all, they have basically broken the system tray — that’s the place where an application I’m developing puts an icon, from which you can use a menu to open a new document or a saved one. This is because it does not fit with their new design philosophy, among other reasons because it was originally intended for notifications, but was taken over by applications like mine that used it for menus, and so different programs use it differently, giving an inconsistent user experience. This is a valid argument, but not for changing something unilaterally and breaking a whole lot of the open source software in existence. It is possible to add an application to a “white list” (which presently includes Skype, the HP printer monitor and Java applications), or to set the white list to simply “all”, but even that does not let my app use it.

If they really wanted to offer a Mac-like experience, of course, they could allow my system tray menu to operate from the launcher, as you can do with one line of code on the Mac. But no, they actually want me to completely re-write my program so that it fits in with their design philosophy, which they have spelled out in one lengthy article after another in Linux Format but which is really unconvincing to the rest of us. My app works on every platform except Unity, which will account for only a small proportion of the desktops out there when the official launch happens this week. Canonical cannot tell the entire Linux-using public (such as it is!) to “unite but follow us”; they are more likely to kill the Linux desktop by doing this than make everyone use the most impoverished desktop environment available.

One of them is the annoyance of the way software updates work. You boot up for the first time in a few weeks and it hasn’t finished updating. Then the anti-virus wants to update. Meanwhile, more updates get downloaded and installed in the background, and after it finishes, a little pop-up appears saying it needs to reboot. It gives you a choice of postponing it for fifteen minutes, an hour or four, but “just let me get on with my work” isn’t an option. On top of that, the anti-virus wants to reboot your machine for itself after having finished a database update.

And two different third-party update programs are nagging you to install new software. I’ve got Safari, which I never use, but the Apple updater also wanted me to install iTunes (even though I’ve never used it on this machine).

This morning, I was helping my aunt do an online job application. We had two hours to do it before closing time at 12 noon today. So, while all this was going on, I told her she might as well make that coffee she said she wanted.

Reboot done, we got down to entering all the details (and the online form is different from the London Borough of Sutton’s own printed application forms, but that’s another story). We got three pages into her five- or six-page supporting statement when a little pop-up appeared and then disappeared. I didn’t see what it was as I had my eyes on the page I was copying. Then a second later, Firefox just disappeared and the machine rebooted. Of course, when Firefox restarted after the reboot, the entire contents of the box containing the supporting statement were missing.

We finally got the statement submitted, less than two minutes before the deadline. But incidents like this are why I simply don’t use Windows for most of what I use my computer for. Ubuntu has one system that does updates (called APT), it never pressures you to reboot in the middle of your work, it doesn’t take ages to let you use anything when the desktop has appeared (because it’s actually still loading stuff), and I’ve never had the thing spontaneously reboot while I’m working.

Some might say that Windows 7 remedies some of these problems, but I’ve never used it and can’t afford to upgrade either of mine. If it does, it should have been a free upgrade given that Vista has never been more reliable than an alpha test release.

Only problem was, when I switched to Ubuntu on this machine, it downgraded me to version 2 again, and when I switched again, to openSUSE v11.2, I got upgraded back. And when that happened, Thunderbird 3 seemed broken - I couldn’t adjust the standard font because the entire preferences window appeared to be broken. It just came up as a blank window. I tried downloading the source code and compiling it myself, but the problem persisted.

I reported it as a bug to Mozilla, and it was eventually suggested that my profile was the problem. I just had to rename the folder containing my old profile, and start Thunderbird (or Shredder, the working title for version 3 which appears when you self-compile). And it worked fine. Clearly the yo-yo upgrading had corrupted my profile, so if anyone’s having problems, that’s the first thing you need to do.

Also, anyone who’s been following my tweets closely knows that I’ve kept my Android phone which I said I was going to send back last month. What the commenters recommended (basically installing Taskiller, which knocks off programs which hog the processor) did the trick, and I didn’t have to install a custom version of the operating system. Besides, I was too addicted to being able to read my friends’ tweets, moderate my blog comments (using the wp2Go app) and even read the odd website on the go.

The reason some folks want us to stop using Mono is simple: they want Fauxpen Source rather than Open Source. That is, they are happy for you to use open source, as long as it’s open source they approve of. Every time — every time — we publish an article about Mono, people write in telling us to stop using it. Well, here’s the news, folks: it’s not free software if you’re forcing your beliefs on other people. … If you want to use Adobe Flash rather than Gnash (an open-source implementation of Flash which isn’t complete), go for it. It’s free software — free as in freedom — and I’m not going to presume that my definition of freedom should stomp over yours.

For anyone who doesn’t know what these terms mean, free software is commonly used (usually capitalised) to mean software distributed as, or at least with access to, source code, so that it can be redistributed and modified, and that modified versions can be redistributed. Free as in freedom means free in that sense, rather than as a zero-cost download. Adobe Flash is not “free as in freedom”, it’s just free to download. Free Software is often called “open source” software, although the terms are associated with different philosophies.

I don’t have a particular bee in my bonnet about these issues, although I prefer good open-source software to good proprietary software, because the latter is less well-scrutinised and is more likely to be buggy. Compare the exasperating Windows update system, which interrupts you five minutes after you’re desktop has set itself up to tell you that it’s just finished installing updates and it really must reboot now, with the relatively smooth update manager in Debian and Ubuntu, for example. However, I fail to see how the editor of one of the world’s major Linux magazines can write such nonsense.

As for why people object to Mono, the reason is not just because it’s based on something originally developed by “Micro$haft”. The reason is that there have been concerns about “submarine patents”, in other words, hidden, patented technologies which Microsoft could use later to destabilise the Linux platform by making patent claims against important Linux applications. I heard this concern expressed on the old LUGRadio podcast back in 2005 (I think it may have been this show). Another reason is that some people regard time and energy spent in reimplementing Microsoft’s technology as a waste when Microsoft will always be one step ahead as they are the inventors of the software, while Linux has its own technologies both for network and desktop development.

It’s true, there are those who want everyone to use free software all the time and talk of the proprietary solutions as “temptations”, when in fact not everyone has the ability to go away and write their own replacement. I certainly don’t. This isn’t the only reason, however, why people object to Mono, and neither is blind anti-Microsoft sentiment, although this may be behind a lot of the emails Hudson receives. There are serious debates about the value of this software, although five years down the line from the first release of Mono, the patent issues that people were worried about have yet to materialise.

For one thing, it didn’t work properly with my graphics card. I had specified an ATi graphics card when I got my Dell in January this year. This is because I wanted a graphics card with dedicated memory, and I’d read that the cards produced by ATi’s main competitor, nVidia, did not work well with KDE at that time. So it was a bit of a disappointment to find that ATi’s drivers for Linux weren’t working with the latest version of the Linux kernel, and that kept up even after the release notes said that they now did work. Fedora’s desktop still worked, but you couldn’t enable the desktop effects; but in any case, I wasn’t making full use of my computer. This version of Ubuntu fixes that.

But I had other annoyances. Fedora’s package management, particularly the desktop front-ends for it, isn’t a patch on Ubuntu’s and it kept reminding me to do updates even when I clicked the “not now” button. It doesn’t tell you that you have to click “Select all updates” before clicking “OK” to actually do the upgrade. Sometimes, when you do a big upgrade, you got little windows popping up saying you needed to restart, except it did this at the beginning of the upgrade rather than the end (so if you honoured it, it would have interrupted the upgrade) and it did one box for every upgraded package that required a restart. Their KDE was also a bit flaky; when shutting down, it always told you the Plasma workspace manager had crashed, and didn’t remember where you put your file management widgets. Linux Format reported a couple of issues back that the response to Fedora 11 had been underwhelming, and I can understand why. It’s flashy at first, but there are a lot of niggling problems.

Installing Ubuntu wasn’t entirely smooth. I always use the text-based installer, which could use an awful lot of TLC - it’s several years old, and it shows. It failed to import my old user account properly, which meant I had to tell it that the contents of my home folder from Fedora were mine (SUSE has done this automatically for years). When I installed KDE and used Firefox with it, the sound wouldn’t work in Flash videos (solution: open up Konsole, type ‘alsamixer’, then press “M” so that the “MM” at the bottom of the slider reads “OO”). Finally, the terminal in GNOME didn’t seem as responsive as it should do. (But I was going to install KDE anyway, which has its own terminal.)

I have a couple of bugbears regarding the package management as well. APT has always been more fun to use than YUM in my experience, but none of the packages were signed, or at least not as far as the system could tell, which meant that every time I specified a package to install in Synaptic, it asked me whether I really wanted to install an unauthenticated package even though it was from Ubuntu’s own repository. When I installed KDE, it interrupted me mid-install to ask me whether I wanted to use the KDE or GNOME login program. It should ask me all the questions at the beginning, not in the middle of a big job.

But now that KDE’s installed and the sound problems are fixed, my computer is a joy to use again in a way that it hasn’t been for months. It starts up fast (shorter booting times are always on the agenda when engineering a new release) and things work smoothly and snappily. When I was out in Kingston earlier, I skipped a trip to the Apple Store to check my email and blog comments so that I could get home and use my own computer — the first time I’ve ever done that.

The reason I ordered it is because the built-in network interface didn’t play nicely with Linux, which is a shame as Dell is one of the few companies which don’t dishonour their warranty if you install it, which is why I ordered a Dell (and they were cheapest). Almost every time I booted up, I would have to use the little network icon to turn networking off and back on again, and if that did not work, I had to use a terminal program, and disable and then re-enable the kernel module that drives the network. There seemed to be no rhyme or reason and how long it would take to get it working was totally unpredictable. The network card cost £16 including VAT and delivery, and when I booted up after installing it, it just worked, no hassle. The machine is a Dell Inspiron 530, which has since been replaced with the Inspiron 537.

In short, I’ve spent the last week and a bit trying one Linux distribution after another, trying to find one that “just works”. The most recent (probably latest but one by the time you read this) edition of Linux Format declares that Fedora 10 “kicks Ubuntu’s ass”, Ubuntu being the distribution bankrolled by Mark “first African in space” Shuttleworth (the name means “humanity” in a South African native language). Ubuntu is the most popular version, and comes in two currently popular versions: the long-term support (meaning three-year support) version released in April 2008, nicknamed “Hardy Heron”, and the 18-month supported version released in October, codenamed “Intrepid Ibex”. Fedora is descended from the old Red Hat Linux, and Red Hat has concentrated on building an enterprise version of Linux and selling support contracts (the red hat in their logo is a fedora). I had Intrepid already installed on my laptop, and I did not really want another system running Ubuntu.

So, after several failed attempts to use Firefox to download the DVD of Fedora 10 the weekend before last, I finally got the whole thing downloaded with a command-line download program (called wget), burned it to DVD and started on installing it. I knew there were problems pretty early on, because I had problems getting the network started; during the boot-up process, it would hang for ages trying to get a network address, and usually fail. This is really unusual; the usual rule of thumb is that if you want to make your network flawlessly with Linux, get a router and connect your computer to it with an ethernet cable. Not this time. A few enquiries revealed that Dell’s motherboards (made by Foxconn) have a history of problems operating with Linux, particularly with the power management feature (Linux has solved this in recent versions by identifying as Windows, much as some web browsers do), but I soon discovered that my computer may have been using the wrong driver, and disabling one and enabling another is a pretty simple job.

Another problem with Fedora, which was a bigger pain than the network problem once I found that workaround, was that its font rendering is not that good. Linux uses a program called Freetype to render fonts nicely by smoothing out their edges; the fonts themselves contain special codes telling the font renderers how to do this, but these codes are patented, so American distributors like Red Hat cannot distribute them without paying patent royalties, and since they cannot guarantee how many copies are being distributed as its open source and do not get paid for the copies they distribute, this makes publishing Freetype with the codes turned on is impossible. The patents are not recognised outside the USA, and you can turn them on yourself by rebuilding Freetype, which I did. The problem was that another aspect of the smoothing was broken, resulting in grey text, especially, looking like a riot of colour, and this affected some situations and not others. So, Fedora had to go. It was just a question of what to replace it with.

My first port of call was OpenSUSE, a distro I have used on and off over the years; it was in fact my first serious introduction to Linux. The problem with it was that Fedora had installed a disk partitioning scheme that the SUSE installer couldn’t deal with, and it gave up with a cryptic error message when I tried installing it. The only solution was to transfer all my files to another computer, completely erase all my partitions and install afresh, which I did. The network problems persisted, and (unlike Fedora) it could not install the drivers for my ATi graphics card. So, that had to go as well.

I had also given Ubuntu a try, but neither of the discs I had downloaded would install: the “desktop” disc - a live CD you can install from - would not install because it cannot deal with the partitions Fedora had set up (note to Ubuntu: get LVM enabled on the Desktop disc, as even if it’s too complicated for the common user, people have LVM partitions already set up which they may need to install Ubuntu on), and the “alternate” disc, which did not even load, displaying one error message about devices it couldn’t read from. I gave the pre-release for the upcoming Debian release a try, and to my delight the network “just worked” during installation (but didn’t once the system was actually installed), but I had no joy in getting the graphics card drivers installed.

So, it then occurred to me that, now that I had got rid of the LVM on my hard drive, the Ubuntu desktop disc would work, so I put it in the drive and rebooted the machine. It worked a treat, and before very long I had a working Ubuntu install, so I got to work on installing everything a programmer needs and, of course, the ATi drivers. It worked, although while downloading them, it actually look like the process had crashed, as the progress bar did not fill up and nothing I did with the mouse or keyboard had any effect. But as the process ended up with the drivers installed, with the desktop “bling” effects enabled, and my attempts to cancel the operation failed (and it didn’t just give up halfway when I clicked Cancel), this is basically a user interface bug.

So, in short, Linux Format’s front-page claim that Fedora 10 kicks Ubuntu’s ass is just rubbish - there are reasons why Ubuntu is the number one Linux distribution, which is that things are more likely to “just work” than on pretty much any other system, because it’s free, because it comes on one CD and not five and not a whole DVD, because it installs quickly, because it offers access to the Ubuntu package archives which are massive, because things like proprietary graphics card and wireless drivers install painlessly and you don’t have to recompile Freetype to get code-based rendering, because its package installer is fast, unlike Fedora’s, and for so many other reasons. It’s too late to get my reply into the upcoming edition (it’s due out tomorrow), but I may well write them a letter in time for their next edition.

Oh, and on the subject of my birthday: I and a load of my family went to an excellent Syrian restaurant in Shepherds Bush, called Abu Zaad, which is pretty much right outside Shepherds Bush Market tube station and serves really great kebabs and other regional food - you can find their menu here. Everyone loved it, including a few people who are normally not that adventurous about food. We had been going to the Moroccan Tagine in Notting Hill, but their prices have gone up and management has changed, and to be honest the change did us good. Most of the food is not hot and spicy and I’d recommend it to anyone (and it’s very popular with the local Muslim population).