A mailinglist about dconf and a description of the current concept/design idea

[Philip Van Hoof]

On Thu, 2005-07-28 at 12:10 +0200, Philip Van Hoof wrote:
> Bleh, I hate mailing here ... so here goes nothing[1]
> 
> Only for people who are interested in the implementation and design of
> what we are going to be calling "dconf". 

I'd like to clarify something about how I, personally*, view the project
"dconf" when comparing it with existing projects like KConfig, GConf and
UniConf today.

There's an extremely important difference (today).

And that difference is that projects like UniConf, KConfig and GConf,
are projects that are usable today and for which your contributions will
or might be in use in a near future. So if you are planning to build on
a configuration management infrastructure and you want to see your code
being used (or usable) as quickly as possible, "dconf" is not the
project for you.

The project "dconf" can be viewed as a research project. My plans are to
design and implement an Utopical system and see what happens. A first
release will just be a prototype. And maybe some desktop environment
community might pick it up. Or might only pick up the (new) ideas. Maybe
none will pickup anything from "dconf".

Also note that I never promised nor said dconf will ever be in a usable
state. So don't count on it. Don't reply to the UniConf people: "Oh we
don't want to integrate with you guys because there's some dudes trying
to do DConf". If you want to: do it.

Please note that infrastructures like GConf and KConfig (and UniConf)
fulfill their requirements. They are usable and they do exactly what
they are suppose to do. The project "dconf" is not about trying to prove
that they aren't. The project "dconf" is, in my humble opinion of
course*, about researching what's possible. About researching how to go
to the next level in configuration management infrastructure.

It's not about configuration management today.

I think that the future of "informatics at home" isn't only about the
desktop. I think it will be (more) about electronic (home) devices that
will also need some personalised configuration. The desktop might play a
role, yes. Maybe even an important one. But I'd like to build and
research a configuration infrastructure that can be used in the future.

So it's not just about configuring your E-mail application and browser.

It's about making sure your car and personal data assistant will also
receive your E-mails, IM messages and video conferencing without forcing
the user into configuring hundreds of devices and duplicated settings
(in the Utopical idea that we'll have wireless Internet in all the big
cities of this world sooner or later).

And sure UniConf and GConf and KConfig can do that. But they aren't
doing this today. So the research is also about: why aren't they?

And the answer is probably: because none of them have been designed to
become an "environment neutral" standard. And none of them have the
potential nor willingness by it's developers to become one. Expect,
perhaps, UniConf has the willingness to become such a standard. But
we'll have to see if they also have the potential. And I'm not saying
they don't. I'm not even trying to make it look like that. 


I hope this is now clarified.



* This is probably not how every dconf team member views it.

-- 
Philip Van Hoof, Software Developer @ X-Tend
home: me at pvanhoof dot be
gnome: pvanhoof at gnome dot org
work: vanhoof at x-tend.be
http://www.pvanhoof.be/