You've got the idea, that's what such changes result in
--- deco/baghiraclient.cc Tue Mar 8 17:00:46 2005
+++ deco/baghiraclient.cc Thu Mar 10 22:55:22 2005
@@ -213,7 +213,7 @@
WId root, daddy = 0;
WId *kids = 0L;
uint numKids = 0;
- XQueryTree(qt_xdisplay(), parent->windowId(), &root, &daddy, &kids, &numKids);
+ XQueryTree(qt_xdisplay(), parent->widget()->winId(), &root, &daddy, &kids, &numKids);
XReparentWindow( qt_xdisplay(), winId(), daddy, 0, 0 );
/*if (BaghiraFactory::effect(client->currentStyle, client->isActive()) == Baghira::Brushed) */shape();
IMO the only reason to create the registry is filesystem design aimed toward large files support, not couple-of-bytes entries, which often means resource waste and great slowdown (as it was with FAT).
That would be nice to have a registry-oriented filesystem (mean real UNIX FS) for that. And btw, the registry problem exists in Wine projects for years, they work with one-large-regedit-export format for now, that really eats resources as the only operations on such "registry" is load-all-at-one and save-all-at-once.
/usr/local/share/kde/registry.kreg on /usr/local/registry/kde (regfs, local)