Re: desktop layout patch

["Matthias Clasen" <Matthias Clasen poet de>]

> It does not seem that there is a consensus on the list for this
> property. At least there is a message of Matthias Clasen with
> an objection concerning the use of manager selection.

No. What I said is that multiple pagers will step on each others 
toes *unless* we use a manager selection in the way Havoc wrote in 
his patch. I still think that the idea of letting the pager decide
about the layout will not really work with multiple pagers, since
all but the layout master have to cope with the given layout and
we will end up having to display an ugly visual indication about this
fact (i.e. something transporting the message "Sorry, I have to display
the desktops vertically, since the panel I'm on is vertical although the
actual geometry is horizontal. I can't change this since I'm not the 
layout
master. If you find this confusing, you're right")

> I'm not against this new property (as I say before). But what
> I understand is that it is really useful for GNOME and KDE because
> GNOME and KDE do not support large desktop!
> You want desktops layer for edge switching if I well understand.
> But why do you not use large desktop?  A layout of X=4 and Y=3
> is more or less equivalent (for the user) to NUMBER_OF_DESKTOP=1 and
> DESKTOP_GEOMETRY= X*screen_width, Y*screen_height. Of course with
> this the layout is given for free and it does not seem that there
> are problems about "nominating one pager as the "layout master" 
> (using a manager selection)" (as say Matthias Clasen).
> 
> So now why we have large desktop and multiple desktops?

Because the people working on choice-reduction for Gnome 2 thought that 
this feature can be discussed away by saying that the distinction between
large and multiple desktops is "artificial" and that in order to get the
lost functionality back, you will "only" have to implement windows 
appearing
on multiple desktops. I.e, if a window crosses the edge of two desktops, 
it
should appear in both:

+---------------+---------------+
|               |               |
|          +---------+          |
|          |         |          |
|          +---------+          |
|               |               |
+---------------+---------------+



I don't think this is really well thought out. Just consider how this 
interacts with
 _NET_WM_DESKTOP=0xFFFFFFFF. The only logical way would be the following, 
which obviously
can't be implemented under X.

+---------------+---------------+
|               |               |
|----+     +---------+    +-----|
|    |     |         |    |     |
|----+     +---------+    +-----|
|               |               |
+---------------+---------------+



The bottom line is: I think you are right, this addition spoils the 
internal consistency of the EWMH, 
but it is probably better to let it in nevertheless, since 
interoperability is generally considered more 
important than consistency.

Matthias