DConf and Elektra

[Philip Van Hoof]

On Mon, 2005-06-20 at 14:00 +0200, Magnus Bergman wrote:
>                  ,-----.                      ,-----.
>                  | BE1 |                      | BE1 |
>  ,-----. ,-----./>-----<   ,--------. ,-----./`-----'
>  | App |-| Lib |-| BE2 |-P-| Daemon |-| Lib |
>  `-----' `-----'\>-----<   `--------' `-----'\,-----.
>                  | BE3 |                      | BE3 |
>                  `-----'                      `-----'
> 
> "Lib" is the the same library used by both the application and the
> daemon. "BEx" are the backends. "BE2" is the backend which contacts
> the
> daemon using a specified protocol ("P"). Think of these two scenarios:
> 
> 1) The daemon isn't available but if the application doesn't depend on
> the features provided by the daemon and can access the backends in
> process.
> 
> 2) The application depends heavily on the features provided be the
> daemon and will use it if it's available.
> 
> Note that it's theoretically possible that the a backend for some
> database server (for example) can provide every feature necessary so 


Please wait and see what we will publicise as detail architectural
concept and idea. Expect this in a near future (weeks, not months).

And lets for today interrupt (and effectively shut down) any possibility
for yet another flamewar about the thing we are calling "DConf".

"DConf", the concept, it's implementation, it's developers nor the
people who are subscribed on xdg need another huge flamewar about this.

We do have a very good idea of what we need. We do understand the
requirements. We are working and mind-sharing about possibilities. Both
conceptual, architectural as technical. There are developers being
interested (so yes when I say "we" I do mean "we". It's no longer just
"me"). There are plans to go forward and start the project.

What isn't needed is yet another flamewar that will in the end draw this
"DConf"-thing on the FDO canvas as if it's just pure vaporware and
bullshit.

Nevertheless I'm also not promising a success (I know I'm repeating
myself). I'm just telling you that we do have all the ingredients for a
success. Please don't spoil it now by starting "daemon yes, daemon no"-
flames. We have the technical knowledge to make the right decision about
this. Please trust us.

For most people it's not an interesting topic but rather a simple
implementation detail. Yet a lot people, who often don't have any
technical background or knowledge about what they are talking about, try
to base their criticism on ideas and concepts purely on whatever they
once read or heard about  .. threads and processes and how bad it is
that "ps aux" shows one more line.

For YOUR desktop it's MOST LIKELY not important. It's MOST LIKELY not
interesting. It's MOST LIKELY not going to be faster if you don't
implement it using a daemon process. In contrary, it's probably going to
be amusingly slower.

And discussing (flaming) about it is really scaring potentially
interested and important developers away for any concept that comes near
to the concept of "DConf".

So in effect it's having a destructive nature.

Therefore if you have implementation ideas and you really want to share
them with us, please write them down on that wiki!

Oh you're even free to start a whole new page about your fuNtAsIqUe and
3l1t3 idea. Please do that. Just don't start yet another flamewar about
it here.

Enough is enough.



-- 
Philip Van Hoof, Software Developer @ Cronos
home: me at pvanhoof dot be
gnome: pvanhoof at gnome dot org
work: philip dot vanhoof at cronos dot be
junk: philip dot vanhoof at gmail dot com
http://www.pvanhoof.be/