Lennart Poettering
Lennart Poettering
Looks to me as if the broker kicks us off the bus. Maybe we hit some message limit, because of the many PropertiesChanged signals we generate or so? if you...
> Instead it would be ideal if we could dynamically define the "resources" that will be installed in a "component". This would allow us to add optional features like development...
It seems everything else gets kicked off the bus too, including avahi and so on. Before we start wild guessing, let's have a look at what dbus broker actually does...
I would normally suggest that the four first commits are split out into a separate PR that we can merge early, since they seem relatively independent of the remaining ones,...
hmm, so you are saying the issue is entirely limited to pid1 and its communication with the bus?
hmm, so in "process_running()" in sd-dbus.c we'll keep processing the wqueue preferably, and only process the rqueue if the wqueue is empty (or the socket is full). This indeed means...
@msekletar so we actually have this dbus method Subscribe() that clients are supposed to call before we send out any signals. It currently has an effect only for the "API"...
codewise looks ok'ish to me. I'll leave final review to @yuwata who is more at home with this than i am I guess.
Original dt support for this was added in 65c2ad985a8debdf6d7d11fee5b466f280260f4b (#24265)