systray specification...

[Carsten Haitzler (The Rasterman)]

On Fri, 3 Feb 2006 10:18:42 +0100 (MET) christian.loose at hamburg.de babbled:

> -----Urspr$(D+d(Bngliche Nachricht-----
> Von: aseigo at kde.org
> Gesendet: Freitag, 03. Februar 2006 08:55
> An: Carsten Haitzler (The Rasterman) <raster at rasterman.com>
> CC: xdg at lists.freedesktop.org
> Betreff: Re: systray specification...
> 
> On Friday, 03 February 2006 08:55, Aaron wrote:
> > On Friday 03 February 2006 00:48, Carsten Haitzler wrote:
> > > On Thu, 2 Feb 2006 23:53:08 -0700 "Aaron J. Seigo" <aseigo at kde.org>
> > > babbled:
> > > >
> > > > hm. the lack of a widget set (and therefore being able to be a
> > > > bare-bones app) is very enticing. but really only if we aren't using
> > > > X11 because otherwise you may as well have a toolkit around ;)
> > >
> > > i do see your point - though any other means of being a systray app will
> > > require advertising data and likely via some ipc socket - and in the end
> > > we then either reuire dbus (may as well require xlib then anyway - we are
> > > just substituting one fat lib for another) or come up with some other
> > > heavy protocol that one way or another sucks in some big library for
> > > communications (ICE, or one of dozens of other ipc mechanisms) - the only
> > > other sane way i see is putting data in files that are shared and
> > > opened/read. this sounds nice from barebones side of things but loses us
> > > network transparency of a display : ( either way - any mechanism we come
> > > up with that is network transparent like the rest of the display is, will
> > > end up being as complex as using xlib i think- unless of course you have
> > > some suggestion - something i haven't thought of? please braindump! :)
> > 
> > well, it's not so much the size / complexity issue as it is ubiquity of the 
> > mechanism. while apache (to pick an absurd example) certainly will never 
> > write to xlib and require an X connection around, it just may use dbus. the 
> > less absurd example is the linux kernel which already uses dbus for
> > hardware events via hal. imagine if hal were able to export its own systray
> > entry. nice, but not likely not possible if we require xlib.
> 
> I don't know much about this topic at all, but after reading this I thought
> that you could maybe solve this problem with a proxy. 
> 
> So the standard desktop apps use X11 IPC for the systray. While non-X11
> daemons like hal talk to a proxy over dbus that does the communication over
> X11 with the systray for them.
> 
> Just a thought. :-)

technically i would say thats the most desireable mechanism. a back end daemon
that a front end "systray hook" talks to and the systray icon hook generates
the systray icon etc.

> Bye, Christian
> 


-- 
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler)    raster at rasterman.com
$BMg9%B?(B
Tokyo, Japan ($BEl5~(B $BF|K\(B)