dbus-broker

This might be revisionist since I wasn't around for the fights, but I think the criticism is split on Dbus the implementation and Dbus as a concept. There were quite a few fights around the implementation of D-Bus itself, as dbus-daemon which provided it was slow and with design faults. A few ideas were thrown around, causing the most fights being moving dbus into the kernel, that is kdbus. [probably the most revisionist statement here] I believe the topic basically died with bus1/dbus-broker which is a high-performance drop-in replacement for dbus-daemon, do read its announcement post here. I think back in the day you'd shake your fist at the clouds and complain that things were fine with pure sockets before, but nowadays I don't think about the daemon anymore. At all. (hooray ^^) For the protocol, I hate citing Vaxry but he's the loudest opponent I can remember, he disliked the underdocumentation of the protocol and that it encoded the types in the wire protocol itself. I'm fine with linking someone else with other criticism.