KRO Status Report #3

David Moreno Montero david at
Sat Aug 20 19:44:24 CEST 2005


I think that almost all basic functionality for the KDE Runtime Observer 
is in place, and just now I'm almost exclusively working on tweaking the 
GUI to get all promissed functionality.

Latest additions are a search view (not finished), a fast search over 
the tree (incremental search, firefox-like), and a very basic time slider.

The time slider is just now a static slider (not enabled) with a pause 
button. When the button is pressed then the events are just stored 
internally, but not shown on the GUI. These way some things may be 
ccomplished: the most important to frezze the status of the view so you 
can see exactly how the interconenctions of the objects at that moment 
are. It can also be used to make loading of observed applications a bit 
faster. Next step is to be able to move around the slide, so you can see 
the exact status at any time point.

Also not yet finished is the signal grid, where you can add objects 
(double left click over a object on the tree view, or with the search 
view) and if there are signals among them, they are represented. This 
part is quite hard... I never realized how hard is to represent graphs 
and the signal/slot relations between object, with several edges, is 
quite challenging. Anyway I found some information on google, so maybe I 
will finish with a good representation, and not hte crappy one that is 

Please, if somebody tries it, I would like to ear some feedback, as I 
dont know is the GUI is nice, or even if it works on other setups.

It's very interesting to observe, for example, konqueror, or kro itself, 
and see how the widgets are parented. Also is interesting see how when 
quitting some applications there are still some objects left that where 
not correctly deleted (KDE's Fault?).

Latest code is as always at the SVN tree, at 
/home/kde/trunk/playground/utils/kro, and a tarball is at No deb this 
time, as I'm using now KDE 3.5 unstable (built with konstruct), and the 
dependencies would be wrong.


More information about the Kde-soc mailing list