Choice breeds complexity for Linux desktop
New tricks for old dog
The success of the Everex gPC this month raises once again the possibility that Linux can make inroads into the desktop market. In stock at Walmart, initial sales of the gPC caused panic on a scale comparable to the recent stock market panic. Not only has the gPC sold well - it has also proved popular.
It helps that gPC ticks all the right boxes. It claims to be fashionably "green" - hence the "g". At $199, it is ridiculously inexpensive. And it comes with a pre-packaged Linux-based operating system. Perhaps more importantly, the gPC is a basic, stripped-down device, suited to simple tasks such as word processing, e-mail and acting as a client for online applications. This is all a lot of computer users want.
The "Linux is set to take over the desktop" story is, of course, a seasoned perennial that invariably provokes a strong reaction from the developer community. But despite its acknowledged technical superiority, its success as a base for servers and its position as a low-cost alternative to Microsoft's Windows software, Linux-based desktops have so far failed to attract desktop users in significant numbers. Statistics from W3C and a recent Linux community survey show it running behind even Apple's Mac OS/X with around two per cent of desktops.
The gPC and Dell's decision earlier this year to offer a pre-packaged Unbuntu Linux deal aside, the future for Linux on the desktop still looks quiet - despite hopeful predictions of its success. A recent report from Forrester forecasts 2008 as the year when Vista - not Linux - will start to erode the current dominance of Windows XP. Forrester estimates that about 40 per cent of corporate users in the USA and Europe will have moved to Vista by the end of next year.
There are lots of reasons why Linux has not ousted Windows - many of them nothing to do with technology or logic and more to do with politics. Nevertheless there are still outstanding technical issues - not least is the problem of fragmentation that confuses users and developers alike. A quick scan of a Distrowatch more than illustrates the simple point - yes I know I like Linux - but which version?
It is not only fragmentation. Linux is also in danger of becoming the operating system that "does everything". Linus Torvald noted in a recent interview that you can find Linux everywhere from a supercomputer to a mobile telephone.
While such broad flexibility could be a good thing, and will be welcomed by enthusiasts, it complicates development of Linux in the world of desktop software and makes the idea of a "mass-market" desktop for Linux - that echoes the Windows desktop market - harder to achieve.®
gOS is still a desktop
The emphase about Google apps is just a wise marketing as well as the emphase about iTunes. At a first glance there're three keys explaining this success story:
- targets the hype apps and devices
- still a computer (not a diskless or a thin client)
Nonetheless gOS is a desktop.
"...does not apply to applications By Charles Manning"
What is really funny for your post regarding fragmentation of linux platforms is that almost sure you never used or tried a linux distribution .The so called fragmentation does not exist or exist in some minds .All ~ 200 distros use the same GNU toolkit and in almost 90% of their aspects are identical.You will find that first time when you will study a program source code a tar.gz.In about 10 years of day by day use of linux i never founded a single aplication what does not work in any distro .
Legacy library support is often poor
One thing that's bugged me for a while about Linux is that all the major distros seem to have a policy of only providing current libraries and "legacy" (old) libraries going back only one major release!
With the speed of Linux development, this can mean that installing a Linux a mere 18 months later can mean that some of your older software will at the very least need recompiling or at worst (no source code?) won't work at all.
Although now very much a niche player, HP's UNIX (HP-UX) had it right - if they ever shipped a shared library with an OS release, they would *always* ship that library in future releases, even if only a few apps from 5 or 10 years ago used it (yes, you even get X11R4 shared libraries with the very latest HP-UX).
I think this is one of the big issues for Linux ISVs and why they are indeed hesitant about entering the Linux market. Yes, one workaround is to link against archive libraries, but you'd be surprised how few apps out there do...