is Xfce getting too bloated ?

[Olivier Fourdan]

Hi,

Well, I've never seen any session manager being very fast, I guess it's
due the the very nature of the session managers, as they have to start
apps and wait for feedback or timeout, etc.

This said, I think xfce4-session is pretty good in its class.

Another problem with the menu parsing. We are aware of the problem, but
the standard is not simple.  Brian said that he wants to rewrite part of
the menu stuff, but we are in the beta release cycle, so it's not a good
time for a rewrite now. Fortunately, there is the cache that
dramatically improved things, and if it's still not enough, you can
disable the menu from xfdesktop settings...

So in a nutshell, you are probably right, but things aren't that bad and
will hopefully improve in the 4.4 version (we need to leave some space
for improvements in 4.4, don't we ;) )

Cheers,
Olivier.


On Thu, 2004-10-14 at 11:02 -0400, purslow at sympatico.ca wrote:
> i just ran a test of Xfce 4.1.90 against KDE 3.3.0
> on my Gentoo system with an  AMD XP 2500+  &  512 MB  memory:
> both are set up with  10  desktops (workspaces) & should auto-start
>  2  Konsoles, Mutt, Lynx, Konqueror, Gvim, Gqview, Gkrellm, Kmahjongg ;
> Xfce is also set to launch KDE services (as defined in the settings window),
> which are started automatically by KDE itself.
> 
> after a reboot in each case & timed from entering the start command
> to appearance of a desktop (tested by clicking a workspace switcher button),
> KDE took  30 seconds  to start, while Xfce took  32 seconds .
> 
> it may be of interest that KDE takes longer to close down
> --  12 seconds  against  4 seconds  -- , which suggests
> that KDE may be more careful about storing session-management information.
> 
> also, the Xfce session-manager is unpredictable about what it starts:
> sometimes it doesn't start Konqueror, sometimes not Mutt (only a 3rd Konsole).
> i don't know whether this would also happen with KDE if i used it often.
> both correctly started all applications in the test above.
>