The endless Linux ready for the desktop discussion

Slashdot has a story from a guy giving up on using Linux on the desktop. When looking into it what he is actually giving up on is having a ‘rogue’ linux system seamlessly integrate into a Microsoft environment. The community is getting better all the time at providing technologies that will emulate your Windows software of choice, but unfortunatly Microsoft is not standing still so its chasing a moving target. So unless you have an IT organisation which is interested in truly supporting Linux you will proably always have some pains. On the other hand I do believe that for a lot of corporate environments Linux is ready for their desktop as long as they want Linux on their desktops. Meaning they tune their purchasing strategies and software choices to doing this.

That said I have myself been thinking about the state of the desktop. As I to have been part of this community for a long time now and there do always seem to be a new big hurdle to cross before we reach Desktop nirvana. Sometimes it can be a bit disencouraging. But if I look back to when I started out with Linux and compare the problems then to the problems we face today I realize that we have gotten a lot closer to being where we want to be. A lot of applications simply didn’t not exist on Linux when I started out. In fact the only two advanced desktop applications I can remember being at a level where they fullfilled my needs where Netscape and Gimp. Today there are a host of applications in most major categories. Applications that come to mind are Inkscape, OpenOffice, Evolution, Gaim, Ekiga, Totem and Rhythmbox.
For instance I remember getting my friends to switch to use Yahoo messenger as it was the only IM service I managed to get working under Linux at the time.

Hardware support has gotten a lot better. When I first installed Linux on my laptop back then I spent a full weekend to get basic audio working on it by grabing CVS versions of ALSA and trying to figure out magic options in the alsarc file. It also seemed like the only time hardware got detected under Linux at that point was during install so when I changed hardware I tended to end up re-installing Linux on the machine. That problem could have been avoided had I understood more of how linux worked back then, but today its not an issue as installing new hardware seems rather automated. I also remember spending a lot of time trying to get Linux to be able to read the ‘Joliet’ CD’s I burned under Windows. Linux at the time only supported another standard at the time (which there was only 1 burning application I ever saw which supported under Windows and that one was silly expensive).

CD burning under linux only existed in the form of the command line cdrecord application, which it took my quite some time to figure out.
Today my burning needs are mostly taken care of by Nautilus.

Hardware support is today as back then still an issue. But it is much less of an issue and hardware that is supported tend to be so in a much more ‘real’ way. Back when I started ‘supported’ meant ‘you can get it to work if you spend a few days on it’, today ‘supported’ means it will plug and play.

We have succeeded in creating standarized abstractions or subsystems for most things today. So when someone writes a new driver you don’t need your application to specifically support it anymore. As long as it is written towards this shared interface it will just work. I am sure there are still some subsystems which are not perfect yet in this regard, but lets face it we are currently working on fixing usability issues more often than we are looking into the ‘how can we get this type of thing to work somewhat at all’ kind of issues.

Its like libgimme-codec. We are not trying to solve the problem of ‘how do we get any kind of support for media playback’ anymore, instead we are trying to solve the problem of ‘how can we do this in the most userfriendly way possible, given the constraints we are facing’.

We might not be there 100% yet, but looking a few years back in time do make it clear we are moving forward rapidly. And I don’t think we need that many more major announcement like the recent Novell Peugeot Citroën deal before the hardware makers all realize that the Linux desktop is here to stay and needs to be supported properly.

Article with some quotes from me

Nathan Willis recently interviewed me about the codec shop launch Fluendo did.
The result is on now
. The result was less of an interview and more an article than I expected, but I guess it still explains a few things around our shop launch.


#1 Jörg Müller on 02.02.07 at 06:53

That’s a truly realistic view!

An alternative desktop is ready for 70% of the “average” desktop users (not gamers) as far as their printers and scanners are supported. With Turboprint, the printer support shouldn’t be a big problem, although that software is not for free. And now, I can buy at least the media codecs that I need.

When I add the prices of the three migration packages that I might need (CrossOver, Turboprint and Media Codecs) it’s still less than a third compared to Vista – a real bargain!

Actually the only obstacles to the breakthrough of the Linux desktop in the mainstream in terms of market share are just _some_ of the drivers. I think it’s only a matter of three to four years from now.

#2 Lars Wirzenius on 02.02.07 at 07:03

For past couple of years, I’ve wondered exactly how ready Linux/GNOME is for the “average user”. So I did a little experiment: (sorry for plugging my own stuff).

Conclusion: except for sysadmin stuff, multimedia, and “toys” (Second Life, etc) explicitly made only for Windows, Debian etch is ready. The experiment was with Debian, but since the same software is available everywhere else it’s in no way restricted to Debian.

#3 Stefano Canepa on 02.02.07 at 08:48

I do think that GNU/Linux with gnome or KDE is ready for the desktop of any user who don’t like to play with PC and who does not need legacy application. I also think that to play it’s better to buy a console and that 90% of legacy applications need to be rewritten and can be rewritten to work on GNU/Linux.