Re: Reducing gconf roundtrips

[Philip Van Hoof <spam pvanhoof be>]

On Mon, 2009-04-27 at 10:00 +0100, Michael Meeks wrote:
> Hi Behdad,
> 
> On Mon, 2009-04-27 at 01:16 -0400, Behdad Esfahbod wrote:
> > Following up on the gconf discussion a few weeks ago and with help from 
> > Matthias, I looked into profiling and reducing roundtrips during login.
> > Here is a first report.
> 
> 	I guess the most obvious win for GConf currently would be merging the
> gconf-dbus code to HEAD 

That's a terrible idea in my opinion. I've taken a close look at the
gconf-dbus port's code several months ago and it's pure horror: using
DBus the way CORBA is used, not how DBus should be used.

Let's just let gconf-dbus be what it was supposed to be: a quick 'n
dirty port to DBus to get rid of ORBit-2 as dependency on certain mobile
devices, so basically something nice-to-have for GMAE and the Nokia
stuff.

Let's not try to turn it into something more, because it's really what
it is. You can't make a person pretty by injecting even more meth into
her arms, nor can you make gconf-dbus look pretty that way.

People like Ryan Lorty are working on DConf, GSettings and GVariant to
have a serious replacement for GConf and its DBus port. Let's just do
that instead. Even maintaining an API that bridges GConfClient to
GSettings is a far far better idea.


-- 
Philip Van Hoof, freelance software developer
home: me at pvanhoof dot be 
gnome: pvanhoof at gnome dot org 
http://pvanhoof.be/blog
http://codeminded.be