Re: GConf vs. bonobo-config

[Havoc Pennington <hp redhat com>]

Miguel de Icaza <miguel ximian com> writes: 
> I could make a similar argument about GConf: to be included in GNOME,
> it should use the standard GNOME component architecture.  As you see,
> we wont get anywhere.
> 

I would distinguish these by saying: whether GConf is used affects
sysadmins and users of the desktop in a significant way. Or rather,
whether we have a single config system that meets the requirements I
think it should meet affects sysadmins and users. 

Whether GConf uses Bonobo is completely invisible to users, it is
purely an issue that affects programmers. If I changed it to use
Bonobo today, the only thing users would notice is that the code churn
introduced new bugs.

Of course, in saying that GConf must be used for GNOME apps I'm open
to pragmatic exceptions; I wouldn't fork or exclude Evolution over
this issue. But I would consider it an Evolution bug/misfeature.

However I think it'd be my responsibility to have good admin tools,
etc. for GConf so that users demanded GConf features. That would be
the ideal way to get GConf support in Evolution.

My view is that whenever you see a project on sourceforge whose
feature list is the list of libraries it uses, that project has a 99%
chance of sucking ass. ;-) If the list of features involves
user-visible cool stuff, then it's probably a program worth
downloading. i.e. I am not going to download the Perl/GTK IRC client
that uses Bonobo and GStreamer. I might download an IRC client with
cool text munging features, pluggable something-or-other, and sound
effects. Or then again maybe I wouldn't. ;-)

You see the point. Library usage is not a feature. But a single config
system that meets the requirements for administratibility, that is a
feature. The reason for requiring (ok, strongly suggesting) the use of
GConf is that it is such a system.

Havoc