[nSLUG] Linux fragmentation?
j.maxuel at gmail.com
Thu Mar 7 16:01:05 AST 2013
I would have to agree that Linux's future is in the end of the desktop. I
see Apple limiting their PC offerings (and touting the range of
iPod/iPhone/iPad in its place). Microsoft made a very bold move in turning
the lion's share of Windows into a mobile OS, yet to see if they have
actually shot themselves in the foot with it. Even Ubuntu has pushed Unity
to make the experience more friendly for the tablet user. I guess the
obvious indicator would be to see OSX and iOS coalesce.
But there's a factor about fragmentation (which de Icaza discusses as a
disaster). I moved to Linux (twice) for more than not just wanting to have
to pay for an OS. I am here because of choice. With Windows or OSX,
choice lies in not upgrading to the next version, to avoid an unwanted
feature. However, this can be short lived as newer software versions often
require newer OSes.
I had Windows XP at the time of the second migration. (I only stopped
having Linux in even a dual boot situation when my main PC fried - I
cobbled together another that had a small HDD and a video card /w tuner
that Linux didn't support, forcing me to choose.) I found XP bloated, yet
was on the market for 5 or more years. I kept up with Linux with VM's
when I had the hard drive space to do so, and noted how small the disk
image was, that I figured it was time to repartition.
Choice becomes how much you want to change our experience. The first thing
I did was ditch GNOME as I was a heavy KDE user for years. To have that
many options, fragmentation is never far away. When it came to bloat, I
found that when I added a dual boot to my netbook, Debian Squeeze loaded
faster (by about 2x) than Windows XP.
I still consider fragmentation an issue. I'm faced with what I believe is
GTK3 dialogs opening on my newly upgraded Debian box (to Wheezy) - the
decorations look about 12 years old. I'm not sure what it is causing it
but I think it has to do with the remaining GNOME packages being removed
during the upgrade (I use Xfce and until Debian makes it their default
desktop environment, I will always have an extra thing to consider when
finding a fix).
I think the solution to fragmentation is to create
an environment where everything in it uses the same (software) components.
To not dis-encourage choice, I do not mean accept one product out of the
box across everything, but all configured to your own taste. If you are
introducing someone to Linux, use your favorite distro (or at least the
internet's favorite). Your friend will create their own taste once the
proficiency is there.
The next few years will be interesting if the big desktop OSes continue to
go mobile (and daily functions continue to be switched to online "cloud"
services), leaving the desktop in the dust. I would like to see how people
(particularly power users) will choose.
"One should strive to achieve, not sit in bitter regret."
- Ronan Harris / Mark Jackson
On Thu, Mar 7, 2013 at 1:36 PM, George N. White III <gnwiii at gmail.com>wrote:
> There has been a lot of ill-informed press commentary on the failure of
> Linux to become a mainstream desktop platform, but
> we have some accounts by people who actually use the stuff they write
> How I ended up with Mac <http://tirania.org/blog/archive/2013/Mar-05.html>Posted
> on 05 Mar 2013 <http://tirania.org/blog/archive/2013/Mar-05.html> by
> Miguel de Icaza: "While reading Dave Winer's Why Windows Lost to Mac<http://threads2.scripting.com/2013/march/whyWindowsLostToMac>post, I noticed many parallels with my own experience with Linux and the
> de Icaza mentions fragmentation of linux. At work we have "mission
> critical" apps, some tied to RHEL and others to Ubuntu, so we have to deal
> with both. From the viewpoint of most uses, the vast majority of the tools
> (compilers, editors, LaTeX, R) are available for both platforms, but there
> are differences in support for devices like printers, sound, drawing
> tablets, etc. We rely on some libraries such as netcdf, hdf4 and hdf5, gsl,
> etc. In practice, the distro versions of these libraries are often built
> with different options or linked with buggy versions of other libraries or
> compiled with buggy versions of GNU compilers. We end up having to either
> install 3rd party binary versions of libraries or build them from sources
> to get working versions on both platforms. I'd like to see the 3rd party
> developers (unidata, hdfgroup) work more closely with distro packagers to
> test and refine the libraries. Too often the distro packager is only
> interested in making stuff work in their own work.
> Meanwhile. Linus Torvalds gets a Google "Pixel" chromebook<https://plus.google.com/+LinusTorvalds/posts/dk1aiW4JjHd>.
> He loves the display, and uses the chromebook for email, but finds it too
> limited for his other needs: "For a laptop to be useful to me, I need to
> not just read and write email, I need to be able to do compiles, have my
> own git repositories etc..". The problem is that there are only a few
> people like Torvalds, so unless he wants to build his own laptop he has to
> work with artifacts that can be sold to broad market.
> The current iPad and high-end OS X laptops also have "micro-pixel"
> displays. iPad apps for systems administrators<http://www.infoworld.com/t/mobile-applications/ipad-apps-the-system-admin-441>mentions using VNC/RDP to run apps on servers/desktops remotely, but I
> don't know if these get any advantage from the higher quality displays.
> Google has presumably put some effort into this, and can take advantage of
> the fact that their web apps are designed to display in a browser.
> Meanwhile, RedHat has moved past VNC and RDP and is working on SPICE<http://www.redhat.com/resourcelibrary/articles/rhev-desktops-spice>,
> a remote rendering protocol for "virtual desktops". I tried spice with
> Fedora 18 in a KVM virtual machine, but the current support in X11 degrades
> for screens larger than about 1024x768.
> Maybe the future of linux is the end of desktop/laptop computing. If
> SPICE matures as advertised, we could have (virtual or physical) servers
> (with wired internet and wireless LAN and cooling but no display). So we
> have two potential paths:
> 1. a protocol along the lines fo SPICE that supports legacy X11
> applications and suitable remote micro-pixel display servers on
> commercially viable hardware, or
> 2. new applications designed to display in browsers (see, e.g., Developing
> Web Apps on Linux) <http://www.linuxjournal.com/article/10789>, and
> native browsers running in commercially viable micro-pixel display devices
> How does this relate to linux fragmentation? If we are using commercially
> viable micro-pixel display devices to work with apps running on linux
> "servers", individual users don't have to commit to a particular linux
> distro, but can choose the distro that best supports each app. Many people
> do a crude approximation of this already, using VNC/RDP to run apps across
> a range of distros, either as developers checking that the apps runs
> properly on multiple platforms or a user cherry-picking apps from various
> George N. White III <aa056 at chebucto.ns.ca>
> nSLUG mailing list
> nSLUG at nslug.ns.ca
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the nSLUG