The linked article is a writeup of one of my spare- spare-time experiments. It starts with the UI/interaction model from Plan9/Rio to try something out of the ordinary when it comes to desktop workflow and security.
Looks great. The feature i love is launching a GUI program from a terminal replaces that terminal. I use i3 and would love that feature there.
I wrote a patch for dwm that does exactly that: https://github.com/deadpixi/deadpixi-dwm
If you look through the older commits, there’s a version that uses the X Resource Extension, and is therefore 100% accurate, but it requires xcb to build.
I’d call 100% accurate a bit of a stretch with Xresource1.2 not fixing the fundamental problem of WM_NET_PID atom etc. It’s still asynch-racing pids, and breaks for non-local connections or if you get creative with fork. That aside, with cooperative clients - it’s ‘good enough’, of course.
The beauty of the plan9 approach is that it doesn’t cost any extra server-side resources other than what has already been reserved for the parent terminal, and subsequently, you don’t have another client doing things in a hidden window somewhere. It also chains nicely. The downside, as Rob himself admits, being that it was ultimately too simple.
True. I meant 100% accurate in that for a given window, it either works or doesn’t, versus _NET_WM_PID, which can be inaccurate, set too late, not set at all, set correctly but for a different machine (with hostname guessing), etc.
This looks even better as an advert for Arcan imo.
It’s something I’ll be looking more into once the summer comes.
it’s indeed a bit more photogenic than both awb and durden, even though they are both doing cooler stuff technically speaking. You know where to find me when summer comes :-)