Proposal and RFC: Introducing DAL, the "Desktop Abstraction Layer"

[John (J5) Palmieri]

On Thu, 2005-01-13 at 10:21, Ikke wrote:
> On Thu, 2005-01-13 at 10:14 -0500, John (J5) Palmieri wrote:
> > > 
> > > [1] http://blog.eikke.com/index.php/ikke/2005/01/09/p66
> > > 
> > 
> > Just read that.  A few points.  D-BUS is engineered for security so I
> > don't really agree with that point.  However linking in external
> > dependencies may be a good reason to not have libdbus added in this one
> > special case.  This however is not a reason to create a level of
> > indirection for all applications.  One can simply write a glue daemon to
> > monitor XScreenSaver's communication channel and propagate it to D-BUS.
> I've got that working (see other blog entries).
> 
> > Having all application go through this or even promoting this as the way
> > to send messages to the desktop, in my opinion would be wrong.
> Ok
> 
> > --
> > J5
> 
> So actually, what would be better than DAL: create a spec (like the HAL
> spec) where desktop apps should comply with, and maybe start writing
> patches for apps giving them the possibility to use the spec, hoping
> they'll get merged upstream, and promote the spec so app devs will start
> making their software compliant to them theirselves?


Yes, and you could also create libraries for certain well used
interfaces that would make implementing them a breeze except the
libraries would have to be desktop specific for the most part.  This is
what we are doing for desktop services like the panel.  We are
developing a GnomeService library which implements a common interface
(start, status, etc.) in which the application can use GTK+'s signal
mechanism for hooking up callbacks to the interface.  If they don't hook
up the method we can have a default signal handler so we can hook up
apps in stages.  We plan on going through a lot of the gnome apps and
making them use this interface so at least in the Gnome world D-BUS is
already going to be pervasive.  D-BUS was designed with KDE in mind and
I think if you give them a good reason to use it they will have little
objections.  The Spec will go a long way and I am sure you will find a
lot of people wanting to implement it.  I know if there is an agreed
upon spec I will go through a bunch of apps and add the interfaces
myself.

--
J5