Wayland broken by design?
I had an interesting conversation on IRC with a fellow who claimed to know Daniel Stone. This individual claimed that Wayland is obsolete since now that the dbus kernel interface has materialized all ipc should be done using it rather than the specified wayland protocol.
This is the first time i have stumbled upon this viewpoint, so i am asking how accurate are these views?
Hah, wonder who that was? And no, I don't agree. Just because you can do something, doesn't mean you should: just have a look at the CORBA/SOAP/XML-RPC crazes over the past however many years.
The same thing happened with dbus, where suddenly your entire desktop was being serialised and rammed through the same daemon, which ultimately became a massive bottleneck. It's a great idea for some things, but when you really are just doing point-to-point IPC which you need to be blazing fast, and you have no particular need for things like access control beyond filesystem permissions ... why?
I should add though: that's just my opinion. There's no absolute right and wrong. I mean, liking microkernels is definitely a valid opinion, just not one I'd recommend.