new dev branch stuff

Erik Harrison erikharrison at gmail.com
Wed Jan 19 18:17:56 CET 2005


I'm not Joe Klemmer (nor do I really want to be ;-) ), but if no one
minds, I'd like to weigh in as a user.


On Wed, 19 Jan 2005 16:22:23 +0100, Jasper Huijsmans <jasper at xfce.org> wrote:
> On Wed, Jan 19, 2005 at 03:04:21PM +0100, Benedikt Meurer wrote:
> ...
> >
> >  (a) "just work"
> >  (b) "easy to use"
> >  (c) "lightweight"
> >  (d) "advanced" (this is least important, IMHO)
> >
> 
> Well, these are everyone's goal, even GNOME and KDE. But accents will be
> different. Lightweight is probably the most distinguishing goal for Xfce
> compared to the others.

I think I speak for at least a few people when I ask, "What does
lightweight mean?" Until recently it's been "Whatever Olivier says is
lightweight" which is perfect, until the developer and userbase gets
so large that the clamor from the masses becomes too large for one or
two developers to make every decision. Then you need some guiding
rules.

As a user, here is what lightweight means to me:

  * A small memory and cpu footprint for the default install
  * Feature complete but not feature rich, for the core apps.
  * Generally performant

The implication, in my mind, of this list is that Xfce should worry
less about how much diskspace it takes up than how much memory. For
example, xfmedia is not launched by default, so if it is installed I
don't see any real "bloat" of the desktop. However, MCS fills the same
ecological niche (if better, in some ways, and worse in others) as
D-BUS, and D-BUS is an emerging and technically strong standard. MCS
consumes memory, D-BUS consumes memory. If all the apps I want to run
next year are going to require D-BUS, it is more lightweight to move
to D-BUS. Not to mention that it saves us developer time in the long
run.

Other considerations are - if lots of core apps need to do printing,
it is better to make xfprint4 a wrapper around an Xfce printer widget,
so that we don't spawn lots of child apps, like Mousepad does now.
That gives us the advantage of making out printer widget behave like
XfceFileChooser, and wrap the GTK+ printer widget when it moves out of
gnomeprintui and into the toolkit.

Very commonly used apps need a plugin system - like Xffm, and the
panel. Xffm and the panel ARE the interface. The window manager, text
editor, etcetera probably do not need the overhead.

> 
> > Atleast the first 3 points imply that the core is small and easy to
> > maintain, which isn't the case currently (I think I say that, as I was
> > the one that tried to release the beast!).
> >
> 
> Please, try to make suggestions for change, especially since you have
> first-hand experience with the release.
> 
> > With all the recent changes and ideas popping up, I thought it was time
> > to ask if we still have a goal, or if we simply follow the tradition of
> > KDE and Gnome (which would basicly mean for me, to re-think if its
> > really worth to spend my spare time on Xfce).
> >
> 
> We aren't doing that bad, are we? Mainly because of the limited time available
> to all of us, we have focused only on the core applications. I agree that this
> is a good time to step back and think about how we really want Xfce to work
> and to think of the best way to get there.
> 
> I'm not sure what you mean by following the tradition of GNOME and KDE.
> Improving interoperability with them is an important part of your (a) and (b)
> and because they have several orders of magnitude more developers we will
> surely end up following their lead more often than not. We also need them to
> supply applications that can be used seamlessly together with Xfce.
> 
> So, what exactly can we do to differentiate ourselves from GNOME/KDE,
> specifically in being light-weight?
> 
> * Less integration: separate programs that work well together and can be
>   removed when not wanted.

More generalization - don't have three separate codebases for the
panel, the iconbox, and the taskbar.

> 
> * Less functionality: we have to make decission on how we want Xfce to work
>   and not add too many fancy feature and options. This is very hard to get
>   right.

Xfce is very modular, or tries to be. Feature mismatch means that a
user can replace the application. 90% of featureitis, it seems to me,
comes from lots of users clamoring to make application X into
application Y which they already use. Don't add a feature just because
somebody else has it - some components will be replaced, and that's a
good thing.

> 
> * Less dependencies: only gtk and libxml2. The trend of putting more and more
>   things in gtk from GNOME may be beneficial for us, or not. On one hand it
>   allows us to use their work, on the other it makes it less favorable for us
>   to implement something differently (linking to both implementations).
> 
> So, by all means let's discuss what we want for Xfce and more importantly how
> we are going to get there.
> 
> Maybe we need more process ('rules')? More coordination? Something else? Let's
> discuss this properly.
> 
>         Jasper
> _______________________________________________
> Xfce4-dev mailing list
> Xfce4-dev at xfce.org
> http://lunar-linux.org/mailman/listinfo/xfce4-dev
> 


-- 
CAPS LOCK: ITS LIKE THE CRUISE CONTROL FOR AWESOME
-Erik



More information about the Xfce4-dev mailing list