Re: server components

[Mike Moran <Mike Moran ee ed ac uk>]

Oliver White wrote:
> 
> Hi folks.
> 
[ ... ]
> However, our system is not aimed at desktops, but rather server
> machines. 
[ ... ]
> For this reason, I'd like to see the creation of a Gnome independant
> collection of components for use in as diverse a range of applications
> as possible. 
[ ... ]
> Does this sound sane to everyone?

Sounds like a good idea to me.

This may not be what you were aiming at, but I was thinking of this sort
of thing in the context of daemons like mserver[1] (a daemon to control
a ppp-link). I recently set this system up again, after searching for
alternatives, since mserver seems to be a bit flaky, and it is also no
longer under active development. I was considering using connectd, but
there is no gnome-native client. This is all down to a different
protocol being used. I don't really see what the fundamental difference
is in the protocols used for each re-implementation of this idea. As
well as this, the basic functionality of reference counted connections
is also re-done in each system.

With the above in mind I was wondering if a CORBA server would solve
these problems? The IDL for the server would be fairly simple, only a
few methods are needed for notification of usage of the connection from
client -> server and the state of the connection from server -> client.
Much of the reference-counting semantics could be implemented in a base
class which implemented the server IDL, with virtual methods left as
hooks for an implementation which actually brings up or down the
connection.

Is this the sort of thing you were thinking of?

[1]: http://w3.cpwright.com/mserver/

-- 
Mike Moran, Dept. of Electrical Engineering, The University of Edinburgh
Web: http://www.tardis.ed.ac.uk/~rfc1149/ http://mike.miezekatze.com/
AvantGo: http://mike.miezekatze.com/Lite/
Mike Moran ee ed ac uk   Your  Message  Here   "I hate you, Milkman Dan"