RFC: DBUS & KDE 4

[Maks Orlovich]

>
> FWIW, I don't think it have to be that specific to conclude D-BUS will be
> well maintained. If we assume the complexity and contribution-threshold of
> DBUS is similar to DCOP, it is reasonable to conclude it will be at least
> as well maintained as DCOP.

I one was to replace libICE w/libLYOD (assuming I would finish the server 
bits) I would estimate that DCOP would be around 7000 LoC[1] of C++, server 
included, not counting the IDL tools and the command-line client tool. 
dbus-0.22's dbus/ directory (which does not include the Qt bindings) is 
30,900 LOC of C, although to be fair that includes quite some unit tests.

So one code base would be more than 3 times larger[2], and written in a coding 
style that's quite different from that of a typical KDE application.


[1] With libICE, it'd be 5000 lines of C++, 8000 lines of C. 
[2] To be fair, it's partly because it's using the STL, while D-Bus has its 
own implementation for everything.