Proposal: New module "kdecore"

Cornelius Schumacher schumacher at kde.org
Tue Apr 18 23:35:19 BST 2006


Working on KDE 4 has become quite hard for application developers, because 
there are frequent changes in the framework which require recompilation of 
lots of code. This is especially hard with kdebase because it's a huge 
module, although only parts of it are required to run an application.

Another problem with kdebase is that it contains lots of stuff which is not 
required for KDE applications when they are run in a different desktop 
environment (be it GNOME, MacOS, Windows, Xfce, KDE 4 apps on a KDE 3 
desktop, or whatever), so it's hard to determine what a KDE application 
really requires (or should make us of, or can presume).

To rectify this situation I propose to create a new module "kdecore" in SVN 
which contains those parts of kdebase which are required for KDE applications 
to run correctly independent of the desktop environment the user chooses to 
run the applications on.

One way to see the new module "kdecore" would be as the runtime dependencies 
of a KDE application, whereas kdelibs contains the compile-time dependencies.

A first approximation of what parts of kdebase could be moved to the new 
module "kdecore":
- the most important kcontrol modules which are required to run and configure 
applications correctly (accessibility, colors, fonts, locale, style, maybe 
more), this also requires the kcontrol framework
- l10n (data needed for localization of applications)
- kdeprint (printing framework)
- drkonqi (crash handler)
- khelpcenter (online help viewer)
- some important kioslaves which are expected to be present on all systems, 
like fish, or are required by other components, like man and info.

I think creating this new module would make it easier to develop, distribute 
and deploy KDE applications inside and outside of the KDE core modules. It 
would draw clearer boundaries to what KDE applications actually are, improve 
modularization and specify dependencies in a clearer way. It could also help 
to advance KDE 4 development by making the process of porting applications to 
KDE 4 more light-weight and more fun.

Comments? Questions? Enthusiastic approval?

-- 
Cornelius Schumacher <schumacher at kde.org>




More information about the kde-core-devel mailing list