xfdesktop woes

Brian J. Tarricone bjt23 at cornell.edu
Sat May 29 22:59:19 CEST 2004

Olivier wrote:

>On Sat, 2004-05-29 at 21:55, Brian J. Tarricone wrote:
>>personally, i'm against anything that requires the user to run a 
>>separate instance of apps to get them to run on more than one screen.  i 
>>think that's a big waste of memory, and i'm surprised that you'd 
>>advocate such a path for xfce, which prides itself on having a low 
>Come on, how many people run such a configuration? People who can afford
>such a configuration can afford the price of the RAM.
i really, _really_ hate it when people try to use this as an argument.  
this is why we have bloated, inefficient, poorly-designed software all 
over the place: because it's so easy to say that RAM/CPU upgrades/hard 
drives are cheap.

sure, my design does tend to cater toward a smaller set of users that 
have multiple monitors.  however, i tend to think that the type of user 
that has multiple monitors would be more likely to use xfce (after it 
gets proper, complete multihead support) than something like gnome or 
kde.  gnome and kde are (in general) for the "average" linux user 
(whatever that is).  the absolute-minimalism junkies will continue to 
use open/black/whatever-box.  the people in the middle, who like a DE 
that is fast and doesn't waste resources, and who don't mind a little 
loss in integration and ease of configuration, will use xfce.  i tend to 
think that the people who have multiple monitors fall into the xfce 
category moreso than others.  sure, that's an unfounded, baseless 
statistic - it's just my feeling on the matter.  but i'm not going to 
change my mind on that unless i see stats to the contrary.

just for some real-world numbers:  right now i'm running feb 12th 
xfdesktop CVS, with the menu module backported.  top reports total VM 
size as 13.8MB.  next i take my latest (albeit somewhat-broken) 
xfdesktop version, that has the multihead support with the design i 
mentioned in my last email.   total VM size: 14.1MB.  both are with a 
single desktop image with a gradient.  my design causes a 2% memory 
increase.  in my opinion, that's a perfectly valid tradeoff for design 
simplicity and elegance, as well as eliminating the need to run an 
instance of xfdesktop per screen.  as i also, said, there is room to 
optimise, and i think it's possible to get my design to be almost 
indistinguishable from the original from a memory standpoint.  (of 
course that these numbers are somewhat rough and shouldn't be taken as 

>BTW, the "overhead" of running multiple instances of an application is
>not that much, most modern system share the code in memory when more
>than one instance is running.
i was under the impression that this is the case for shared libraries, 
but not normal static binary code.  please correct me if i'm wrong.

>Speed is more important than memory footprint. Saving memory is not a
>goal by itself, it's interesting because it makes the system feel faster
>(as less memory gets swapped to disk).
in this case "feel faster" = "faster".  hard drives, as secondary 
memory, are dirt slow.  i think any opportunity to reduce memory 
footprint should be taken, if the goal is to decrease swapping.  in the 
most general sense, a low-memory application is faster than high-memory 
application.  your opinion obviously differs from mine, but i don't see 
much of a distinction between "speed" and "memory footprint" when my 
goal is to make something "lightweight".

i like my multihead-xfdesktop design.  i think it's a sound, simple, 
elegant design, and it thus far runs quickly and without a significant 
amount of extra memory.  some tweaking and optimisation may prove to 
reduce that further.  assuming i can resolve the app-crashing issues, i 
intend for xfdesktop 4.2 to use this design, barring any changes to 
xfce-mcs-manager or other xfce backend features that may make another 
design better (in my opinion).  if this is a problem, then.... well, 
then we have a problem.  i don't mean to be confrontational, but i've 
put a lot of work into this, and i feel very strongly that i'm going in 
the right direction.


More information about the Xfce4-dev mailing list