Something stupid... but might be possible


Forum: Other Help Topics
Topic: Something stupid... but might be possible
started by: rius09

Posted by rius09 on Feb. 06 2007,13:23
Well... a couple weeks ago... I had an idea that idea was....


To install KDE or XFCE or GNOME on DSL and make it work...

Now my question is... is it possible?... and if it is... can somebody do it?

Posted by mikshaw on Feb. 06 2007,17:22
It's possible, yes.  I think your last question is the big one: can somebody do it?
The answer to this is, of course, yes. However, one must consider whether somebody actually *wants* to do this, if somebody has any desire to increase the size of DSL by 500% (a wild guess...i didn't really feel like figuring out an accurate estimate), as well as increase the necessary system resources needed in order to run something like KDE or Gnome. Both of those environments are in direct contrast to the very most basic concept behind DSL's creation.
I'm not saying don't do it, but that most people just have no desire, considering KDE is already available practically every other distro.

Xfce is a little different story.  As far as I know there are people who have gotten it installed to DSL through apt-get.  I personally had started once to build a mydsl extension of xfce, but discovered that it is also a very huge package itself, and not worth the effort (in my opinion).
Of the three, Xfce is the most likely candidate for a DE in DSL.

Posted by torp on Feb. 07 2007,03:33
adding kde to dsl.....jeez just use knoppix and save the heartburn for children, grandkids, college tuition or whatever....

torp

Posted by lucky13 on Feb. 07 2007,09:56
Quote (mikshaw @ Feb. 06 2007,12:22)
It's possible, yes.  I think your last question is the big one: can somebody do it?
The answer to this is, of course, yes. However, one must consider whether somebody actually *wants* to do this, if somebody has any desire to increase the size of DSL by 500% (a wild guess...i didn't really feel like figuring out an accurate estimate), as well as increase the necessary system resources needed in order to run something like KDE or Gnome....

How big would it be? It would depend entirely on which parts were installed. KDE and Gnome and Xfce consist of various parts. It's still K when someone chooses Open Office or Abiword/Gnumeric over KOffice or does away with the Kicker, KEdit, Kate, KAlarm, KMail or other K-items. Each of those DEs can conceivably stripped down so that it's still quite functional, though perhaps not nearly enough to fit in with the DSL mindset. Stripping out apps probably also doesn't fit in with the principle of a unified set of programs and libraries that undergirds projects like KDE, Gnome, and Xfce, but most users will add heterogeneous applications and their libraries to KDE, Gnome, and Xfce systems (especially the latter) anyway.

Here's an example, though, of what can be done to reduce the size of a desktop environment. < KDEmod > is a modular KDE set-up for Arch Linux. I agree it would be ambitious and self-defeating (not to mention masochistic) to tackle something like that for DSL. It would probably be easier to just use another distro if someone really wants K or Gnome or Xfce, even in a "light" manner.

One that comes to mind is < Featherweight Linux >, a Feather re-master which uses KDE 3.3.2. The guy says his ISO is about 240 MB and 700 MB on the hard drive. A lot of that, though, is from all the other stuff he added (Mozilla Suite, JRE, Shockwave, Lua, etc.) in addition to K.

That said, and this is a big peeve (especially when I hear or read people complain), I think there's something to be said for using the same set of shared libraries in reducing total system storage and memory footprint. The libs and apps in KDE are shared and designed to work together. I've used K (wrong tense, I'm using it now on my Kubuntu box) and it's a full suite of apps that works together seamlessly. I love it, it's fantastic, it genuinely makes life easier. Nothing against Gnome or Xfce; I've used them both and they're nice, but they lack the fit and finish and interoperability of all the parts.

As to the claims about how bloated it is, compared to what? Comparing fluxbox to KDE is comparing apples and oranges. Even beyond the eye candy, KDE is a lot more than the sum of fluxbox, sylpheed, dillo, emelfm, and a handful of mydsl apps. It's all of that wrapped up in one fully-integrated package with shared libs. Those libs may be larger than similar groups of programs that accomplish the same tasks, but the whole is less than the sum of its parts when it comes to system resource use.

< This analysis >, and I concede it was done by a KDE supporter who openly admits his bias, shows that the smaller initial footprint doesn't necessarily equate to the smaller overall footprint (and vice versa). His test shows that KDE's ability to reuse the same libs when running apps keeps its memory use smaller than a window manager or another DE that doesn't share libs among native apps:
Quote
Although our libraries cause us some overhead, like the initial large requirements, they are our advantage. They allow us to write good applications that are often more lightweight than competition and of course there are many other benefits like large code reuse, many features and so on.


The worst performer in his test was Xfce running third party apps, because it lacks native apps. Xfce might seem a good choice on the surface for a small-footprint system because its shared libs are smaller, but not when you add all the other stuff on top. That's where the rubber meets the road!

And that's one of the shortcomings, imo, of the mydsl extensions. They don't share a continuity of reusing a unified set of libs, which increases footprint (bloat!).

How many myDSL apps "share" the same non-DSL (ISO) libs? I think the GTK1 version of Abiword comes with GTK2 libs so it's functional in DSL-sans-GTK2. I think audacity does that, too. If so, there's a quick example of redundancy (bloat).

While I appreciate DSL's modularity, there's a point of diminishing return when a user keeps adding back things that were stripped out -- and more so when the apps being added to DSL use completely different libs. You're not exactly reducing bloat, or encouraging reduction, when you're not sharing libs and end up duplicating things (which fits my definition of bloat) because app A requires one bunch of libs and app B requires another set. Some extensions use Gnome libs, some use K/qt libs. I love k3b, but the mydsl package for it is virtually the same size as the DSL ISO!

So how big would a minimalist KDE set-up have to be to give the same functionality as...
DSL ISO:    50  M
K3B:           48  M
OpenOffice: 99 M
Skype:        11 M
MozSuite:    14 M
Rox:             1 M
Grip:             1.5 M
audacity:        5 M
gtkam:            2 M
zine:             5.5 M

total: Well, that's pretty dang close in size to what Featherweight Linux is as an ISO. And that's with KDE "bloat," but one man's bloat is another's functionality -- especially when comparing a full-fledged desktop environment to a window manager like fluxbox.

How big is your DSL once you dress it all up and make it functional for you? Maybe it's big enough to warrant tying things together with reusable libs and apps designed to work together -- like KDE -- instead of throwing together a hodgepodge of crap that's redundant, doesn't work together, and results in the very bloat you despise?

"Increase the necessary system resources needed in order to run something like KDE or Gnome," indeed, or just increase them with mydsl extensions. The choice is yours.

Posted by mikshaw on Feb. 07 2007,12:18
I can't disagree with much that you've said here.
However, in order to benefit from KDE's shared libraries, you must also be using KDE applications. For the most part, I do not. Even on my Slackware system I do not have KDE installed, because it's a waste of space and resources when I don't use the KDE environment and when there are only a few Qt applications that I've found to be a better alternative to smaller, lighter apps built to use gtk1, fltk, xlib, tk, or the terminal. For those few KDE applications, I've simply decided to take the loss. Things like Hydrogen and Rosegarden would be great fun to use, but instead I settle for Trommler and Audacity (Ardour is just too complicated for me).

I also agree that there is redundancy in many mydsl packages, but in some cases I consider it a useable tradeoff for the modularity. In other cases I simply do not use those extensions. The K3B extension, for example, is a fat cow, and I prefer gcombust anyway =o).
The extensions I've built myself have for the most part tried to minimize redundancy and use DSL libs whenever possible, because I usually compile them against the libs in DSL. The Audacity extension does not use redundant libs, by the way...it was compiled against gtk1.2. It does use wx widgets, however, which is something that could not be avoided.  Future versions may need to be compiled against gtk2, but I'll deal with a solution for that when the time comes.

My bottom line with respect to KDE is still that it is much fatter than it needs to be, and my personal uses show it to be worse than using lighter alternatives. If I do not use the KDE environment, K apps take a terrible time to load, and often seem to be unstable (possibly a result of not using the desktop environment?). If I use the KDE environment, I see a startling increase in memory use and a minor decrease in application response.  What it comes down to, I guess, is what applications you use.  If you use a lot of K apps, KDE is probably a better choice. I do not, and in fact tend to have only one or two GUI applications opened at any one time, so my computer's resources are better reserved for those times when an application needs them (running Unreal Tournament, playing movies, etc)

Powered by Ikonboard 3.1.2a
Ikonboard © 2001 Jarvis Entertainment Group, Inc.