@bruns, is this good to go now?
Shouldn't we make sure that entries get removed from the config file properly?
I like very much your proposal about using two classes to avoid reintroducing problems in the future.
Something like this https://phabricator.kde.org/P439
Now HAVE_DBUSMENUQT won't be set. Previously it was set to 0 in this case
Hm. I've had a quick look and I think the intent was that only the slavebase overridden functions call error or finished. So technically all internal functions that can have an error need to set an iError or return one so that the "public" function can then issue error() as needed. This seems to not very well enforced and probably never was, in fact there's a comment in the .h about this very fact
- Refactor the naming of components etc a touch (and hide private bits)
- Fix a few bits of stray naming in the docs
Looks good to me. The slave indeed must issue an exit state there.
Remove dead code
Add context about filename #
Do I have to reset something on my machine after this?
Given it's brokeness and relative usefulness I propose to remove it.
The commit log says "Breeze now knows it's a KMessageWidget". Ok. And now? The commit log lacks what issue this patch fixes...
kdialog also uses KPassivePopup
The case where no notification daemon is running is IMHO an edge case that we don't need to support.
I need this additionally
Fix email in copyright declaration