Yet Another reason to use linux...

Tue Apr 3 22:51:28 UTC 2007

On Tue, Apr 03, 2007 at 05:28:55PM -0400, CLIFFORD ILKAY wrote:
> I'm a long-time KDE user and love everything about it but its resource 
> utilization. KDE and the facilities and tools it provides are 
> ingrained in my development process. I also love my trusty old 
> ThinkPad A21p. I'm not fussed by the fact that it's a P3/850 or that 
> some of the markings on the keyboard have worn off. I'm a touch 
> typist and have developed muscle memory for which keys I need to 
> push. The only things I find lacking on this machine are RAM and disk 
> space. The latter is easy to address for a very nominal amount but 
> the 512M of RAM I have is already maxed out so that limits the 
> continuing usefulness of this machine as a developer's platform.
> 
> In what turned out to be a futile attempt at finding a reasonable 
> alternative to KDE as a desktop manager while still continuing to use 
> the KDE tools that I depend on, I tested fluxbox and XFce. I'm 
> running Fedora Core 6 on this machine. It's a piggy distro and for 
> the most part, I don't really like it but I don't have the time to go 
> on another round of testing different distros right now. I already 
> run Etch, Mandriva, CentOS, and Ubuntu (server-only) on other 
> machines and I've run Kubuntu, OpenSuse, Mandriva, and Gentoo on this 
> laptop so I've already evaluated a few alternatives and have found 
> them wanting in one area or another. I don't find that APT has any 
> clear advantage over RPM, despite the hype around it but that's a 
> different discussion, which I'll leave for another day.

If you use the kde tools you will get the resource requirements of KDE.
The window manager part is actually not a big deal resource wise, the
kde subsystem that all the tools use for communications is.  So pretty
much any kde tool will cause a resource problem.

As for apt versus rpm, well that's not really the thing that makes
debian better.  It's a nifty tool, it managed the downloads and upgrades
for you and resolving the dependancies, something which took years for
rpm based systems to gain.  The real reason it is better is that it's
packages actually have accurate dependancy information, and are put
together consistently and in a way that everything can work together and
get along.  It doesn't matter if your packaging tools work well if the
packages they are managing are crap.

I just upgraded my 486 which I installed Debian 2.1 on in 1999.  It is
now upgrading to 4.0 (ok so I am doing it slightly early), and other
than having to purge a couple of ancient packages for netscape 4.77 I
forgot I had installed (which were getting in the way of upgrading to
X.org it told me), the upgrade has gone just fine (the power failure in
the middle was a bit inconvinient, but the upgrade system doesn't mind
it would seem.)  It has gone through 2.1, 2.2, 3.0, 3.1 and now 4.0.
Gotta love quality packaging of software, by people that do it just
because they care, and hence release when it is ready and not a moment
before.

> Empirical testing of fluxbox, XFce, and KDE on my laptop contradicts 
> the folklore of XFce being a lightweight alternative to KDE or Gnome. 
> Speaking of Gnome, I couldn't be bothered testing it because I knew 
> I'd never use it and would have to remove all the Gnome junk after 
> installing it just for testing anyway. Besides, in previous tests of 
> Gnome within the last couple of years, Gnome's memory footprint was 
> indeed lower than KDE's but it wasn't enough to compensate for its 
> ugly UI and its poor attention to usability. If I have to put up with 
> that, I'll go the whole way and just use something like fluxbox.

I actually find the current version of gnome much better than they were
in the past, and have actually ditched kde in favour of it.  Now using a
different window manager is no help as long as you continue using
anything kde related.

> My testing methods were as follows. I ran four tests, the first with 
> no X running (init 3), the second with fluxbox, the third with XFce, 
> and the last with KDE. Between each test, I rebooted the machine to 
> clear any potential after-effects of the desktop manager in question 
> not cleaning up properly after itself. The same daemons were running 
> in all cases. Only in the KDE case was I running any applets, such as 
> knewsticker, kalarm, kgpg, klipper, and a host of others. If 
> anything, the KDE memory consumption would be lower if I didn't run 
> those applets so in effect, I was discriminating against KDE, my 
> favourite. These tests were done in mid-February, 2007 with the 
> latest versions of the respective desktop managers from the Fedora 
> repositories. 
> 
> I've pasted the results of the "free -m" command for the four test 
> cases below. I've modified the spacing on the first two rows in the 
> hope that the formatting won't be screwed up by line wrapping.
> 
> My baseline - console only
> --------------------------
>         total   used    free    shared  buffers cached
> Mem:      502    212     290         0       14    152
> -/+ buffers/cache:         44        457
> Swap:         2000          0       2000
> 
> fluxbox
> -------
>         total   used    free    shared  buffers cached
> Mem:      502    363     138         0       29    272
> -/+ buffers/cache:         61        440
> Swap:         2000          0       2000
> 
> XFce
> ----
>         total   used    free    shared  buffers cached
> Mem:      502    497       5         0       36    322
> -/+ buffers/cache:        138        363
> Swap:         2000          0       2000
> 
> KDE
> ---
>         total   used    free    shared  buffers cached
> Mem:      502    445      57         0       33    310
> -/+ buffers/cache:        101        401
> Swap:         2000          0       2000
> 
> I ended up sticking with KDE because the difference in memory 
> consumption was not worth the loss of productivity and inconvenience 
> of switching to fluxbox and there were no gains to be had using XFce 
> anyway.
> 
> I simply don't believe Xubuntu would be very pleasant to use on a 96M 
> or 128M system. My son is running that very distro on a Celeron 366 
> with 196M of RAM and while it is barely useable, it is slower than 
> when that same machine was running Windows 98. I think we do Linux a 
> disservice when we make these claims that are not only unsupported 
> but actually contradicted by the evidence.

Well Linux is not to blame when people choose to run bloated
applications on it.  I don't blame windows for some of the java
monstrosities I have encountered.

Most office applications, graphical browsers, and such are simply
amazingly inefficient on resources.

--
Len Sorensen
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists