This isn't my workflow but my friends. One of my friends doesn't use plasma but KDE with plasma-windowed plasma-nm, but there's a piece of code prevent him using plasma-nm. See code here. https://projects.kde.org/projects/kde/workspace/plasma-nm/repository/revisions/master/entry/kded/service.cpp IMHO it's so wrong to detect plasma-shell/plasma-desktop dbus name ( Think about we also have plasma-netbook in kde4). There're quite a lot of better way to do this, if the purpose is to avoid running this module in other DE, for example, detect KDE_FULL_SESSION environment variable, or use kded LoadModule interface to implement lazy initialization by making a load module request from applet.
Git commit 5294e9f0087735abb801b24c8d7bc69a952a5848 by Jan Grulich. Committed on 18/07/2014 at 06:26. Pushed by grulich into branch '0.9.3'. Comment out code delaying initialization of our secret agent until we found a better solution Related: bug 333905, bug 335754 M +8 -8 kded/service.cpp http://commits.kde.org/plasma-nm/5294e9f0087735abb801b24c8d7bc69a952a5848
Git commit 9edd166c9abfbc7c834fba4fed1c730289580261 by Jan Grulich. Committed on 21/07/2014 at 11:22. Pushed by grulich into branch '0.9.3'. Properly delay initialization of our kded module and drop support for plaintext files REVIEW:119353 Related: bug 333905, bug 335754 M +0 -11 editor/connectioneditor.cpp M +11 -39 kded/secretagent.cpp M +4 -4 kded/secretagent.h M +25 -20 kded/service.cpp M +5 -3 kded/service.h M +3 -1 libs/editor/connectiondetaileditor.cpp M +2 -0 libs/handler.h http://commits.kde.org/plasma-nm/9edd166c9abfbc7c834fba4fed1c730289580261