I agree with you. As I said before, I've already been implementing some of
this stuff for a side project. I got the window grouping stuff wroking last
night and "Sven's" technique I had already partially implemented the day
before (mostly works except no event handling). Below is how I am expecting
my prototype to work (psuedo code):

/* the "window manager" creates 8 surfaces */
fb.CreateSurface(&top);
fb.CreateSurface(&topright);
fb.CreateSurface(&topleft);
fb.CreateSurface(&right);
fb.CreateSurface(&left);
fb.CreateSurface(&bottom);
fb.CreateSurface(&bottomright);
fb.CreateSurface(&bottomleft);

/* the wm draws its border into these surfaces (not shown) */

/* the wm sets these via some function in the layer */
layer.setBorderSurface(top, DFWD_TOP);
layer.SetBorderSurface(topright, DFWD_TOPRIGHT);
/* ...and so on for the other six... */

/* then the wm creates an event buffer for events from the border regions */
layer.CreateBorderEventBuffer(&eventbuffer);


Does this make sense? Other things that I believe will need to be added is
the ability for the window to "hint" that it wants the decoration (just
another window option?) and also a way to store and retrieve a "caption" in
a window. The wm would drawString the caption onto the border. If a window
hinted that it wanted decorations, then it would have four surfaces created
and associated with it (it could be made smart also, so if only a top
surface is set in the layer then only a top surface would be created in the
window). These four surfaces would be drawn into using the surfaces set on
the layer object (and redawn when the window was resized).

I'll probably have most of this prototyped today. Let me know what you all
think.

Stephen



-----Original Message-----
From: Sven Neumann [mailto:neo@;bender.convergence.de]On Behalf Of Sven
Neumann
Sent: Saturday, October 26, 2002 4:32 AM
To: Stephen Perez
Cc: [EMAIL PROTECTED]
Subject: [directfb-dev] Re: Window Decorations


Hi,

"Stephen Perez" <[EMAIL PROTECTED]> writes:

> Could someone enumerate through the reasons why Hallvar's idea of
> arbitrary window groups is not so good? It is kinda growing on me. I
> think it goes a long way to keeping the DFB interfaces simple.

IMO it doesn't make sense to manage four windows to decorate a fifth
one. Event handling as well as window stack management would become
a nightmare. I might be wrong but IMHO it makes more sense to special
case the decorations so that they can be handled by a single
structure.

> Also, I've seen it asked on these lists several times: "what's next
> for DFB?" How is this decided? On this list?

It's been asked and answered just yesterday on directfb-users.

> Just curious. Here's my two cents.  I hope DFB stays small and
> light, providing hooks for others to implement anything they might
> want on top of it. I hope DFB doesn't become a full-fledged
> windowing system, although I certainly hope there are plenty of
> windowing systems built on DFB. Put another way, I would hate to see
> DFB (the low-level graphics stuff and window stack) become tightly
> coupled to any particular mechanism for managing windows,
> cut-n-paste, etc. I'm just wondering what a road map for DFB looks
> like. Does Convergence have long term requirements?

convergence has the long term requirement of keeping it small and
light but still featureful. convergence doesn't need window
decorations, clipboard, drag'n'drop and all those other goodies a
full-fledged windowing system has.

On the other hand we'd like to make it possible to build a desktop on
top of DirectFB. To achieve this we'll need a decent IPC framework and
we need to provide the hooks for window management and decorations.


Salut, Sven


--
Info:  To unsubscribe send a mail to [EMAIL PROTECTED] with
"unsubscribe directfb-dev" as subject.



-- 
Info:  To unsubscribe send a mail to [EMAIL PROTECTED] with 
"unsubscribe directfb-dev" as subject.

Reply via email to