I have tried using wlroots, it's very hard. If you had to build your own compositor Mutter is far easier (it provides a GUI toolkit and all you need graphic wise). QTWayland is very nice too but won't get you far outside embedded uses (no xdg portals, no screen capture, etc.)
WayFire makes wlroots a bit easier but I find it quite messy/not clear (but it's very powerful & flexible).
There's definitively the need for an easy high level API for wlroots (the main wlroots dev started working on a high level scene tree API some times ago but it now seems kinda abandoned)
Enlightenment has support for Wayland. It is up to the hundreds of X11 window manegers if they want wayland or not. There a lot of distros supporting x11 WM:s very few supporting Wayland, Enlightenment, Sway, Wayfire, Taiwins, Liri shell,Hikari, a few supports Plasma and Gnome. Only Arch supports them all.
Having a singular library is more limiting than X even for a developer of a wm and much more so to user who can no longer merely assemble an environment by setting a list of components to be started by a script.
That's exactly how I'm assembling my Sway session. My Sway config ends with `exec systemctl --user start sway-session.target`, and that target links to all the services I want running in my session (redshift, notification daemon, etc.). If you don't like systemd, you can `exec my-session-startup-script.sh` in the exactly same way.
> User services only really make sense for user stuff that requires more complex set up and tear down than start foo and kill it when I log out.
>
> Creating a user service per single line exec foo is extra ceremony.
Then go with the parent post's other suggestion: a simple startup script that ends with launching Sway.
> I have nvidia hardware so sway makes no sense for me to use.
I mean the fact that Nvidia refuses to play by the rules, making their hardware unusable with Wayland, is a well-established thing by now.
> Also it's developer is extremely abrasive.
Is he? He has strong opinions and doesn't beat around the bush, but does that really matter unless you yourself actually want to develop Sway?
> Is there anyway that sway does better than i3?
I actually haven't tried i3, but on the face of it I would say that it's an advantage of Sway that it can run without the gigantic and increasingly unmaintainable legacy system that is X. You can certainly run X without problems today, but isn't it widely agreed that X doesn't exactly have a bright future? Sway is not tied to X's future.
I asked you to name even one thing sway did better and you tell me about the nature of the legacy code base.
No user on earth cares any more than they care about the sharpness of the blades used to grind the sausage they ate for breakfast.
I'm not sure why you imagine that nvidia is obligated to support Linux in the fashion you would prefer. I bought the hardware based on the existing drivers working well under Linux and windows not love for them or idealogy. This is why most people buy things.
X11s future departure seems like a thin justification seeing as it's future seems pretty secure for the next decade. Maybe by 2030 there will be a compelling case for switching plus hopefully the kinks will have been worked out courtesy of folks like yourself.
> I asked you to name even one thing sway did better and you tell me about the nature of the legacy code base.
OK, let me rephrase it: a feature it has over i3 is a higher probability of running on the platform provided by most distros in 5 years' time.
> No user on earth cares any more than they care about the sharpness of the blades used to grind the sausage they ate for breakfast.
Indeed. But if you're choosing a sausage you wanna have for breakfast for the next few years, you might choose better than the one made from the animal that people fear might go extinct.
> I'm not sure why you imagine that nvidia is obligated to support Linux in the fashion you would prefer.
Oh not the fashion I prefer. The fashion the Linux developers prefer. Most other providers of hardware seem to. What makes Nvidia special?
> X11s future departure seems like a thin justification seeing as it's future seems pretty secure for the next decade.
Perhaps. You may well be proven right, but I wouldn't bet that X11 is able to attract sufficient manpower to keep up going forward. Time will tell.
I'm not trying to convert you. I'm trying to give one argument why Sway might offer something that i3 doesn't.
Give it some time. With the way sway and wlroots are developed, someone will create a proper high level framework-to-write-a-WM one day and we'll get all those experiments on wayland too.
One day. While today it's easy to write one in X and you don't have all the compatibility issues talked about in the article and up and down this thread.
i3, xmonad, elightenment, ctwm, WindowMaker and dozens of others appeared only because X made it easy.
Wayland makes developing window managers difficult again.