just based off release notes arcan always felt like a much more compelling display project than wayland.
wayland: if we move the compositor into the kernel we can solve several performance problems.
arcan: if we make a universal transport layer we can solve most topological problems.
Plus I like arcan because it reads like it is one guy, deciding the world needed a better display server. and just keeps at it, day to day, no team, no consensus, just getting it done.
however disclamer: I made a port of arcan to openbsd and could never quite figure out what to do with it. the curse of a display server with not enough software I guess.
> wayland: if we move the compositor into the kernel we can solve several performance problems.
Where exactly does Wayland introduce a kernel-based compositor?
Existing Wayland implementations just consolidate what was often several userspace processes in the composited X11 model into a single monolithic userspace process responsible for the window management, input multiplexing/routing, and compositing. I'm not certain this is even a strictly specified aspect of Wayland either, it may just be convenient and simpler to implement this way while enjoying some natural performance advantages.
There's no kernel Wayland compositors AFAIK. Perhaps you're conflating KMS with Wayland since they kind of overlapped chronologically? KMS enabled running XOrg without requiring root using the modesetting driver, it's been a huge step towards cleaning up the graphics stack on Linux, independent of Wayland.
You are correct of course. as an openbsd user it turns out I don't actually know that much about the actual advantages of wayland. I tend to look at it from the rather myopic view point of what it is missing rather than what it has
This is entirely unrelated to the discussion, but something interesting happened to this subchain and I can't just wander past :). I came back to this tab after a few hours to find (at the point I'd come back) that all three posts in this chain were marked as being written "45 minutes ago". All three of them. After having a bit of a memory crisis ("I read this last night... didn't I???") I searched for the text in one of them in Algolia, which found the post and confirmed it had been posted hours earlier. Algolia's now resynced to HN, and shows the same timestamps. Didn't properly investigate at the time. Woops. At least I grabbed a couple of screenshots. https://imgur.com/a/fInenj4
> disclamer: I made a port of arcan to openbsd and could never quite figure out what to do with it. the curse of a display server with not enough software I guess.
Port stuff to it?
Out of curiosity, did you base the OpenBSD port on the NetBSD port?
I never published it due to bugs, so shame on me but it was a fresh port just from the sources, the author has a post I enjoyed where he explains some of the differences between how linux and openbsd do things, I sort of took this as a sign that the author at least tried it on openbsd and sure enough it built without too much trouble.
I don't mean to diminish the post or the author, but I spent a decent amount of time reading through the pages and am about to try their code, but I get the feeling the author is a little either crazy or out of touch with how to communicate with people. It is clear they have made something interesting, but I can't help but feel there is something off in the communication of it.
For what it's worth I happened to talk to the author a while back and we incidentally had a perfectly functional and pleasant conversation, he explained Arcan had been used in a variety of different contexts including machine-vision image recognition. I think this is more of a fundamental focal-point issue: this person can see something in what they're building that would only be appreciated by not just fellow engineers, but those specialized in the niche subfield of graphics that Arcan occupies. But that's the problem: I can't figure out what that niche is, and it would seem I'm not the only one. It feels extremely interesting, abstractly speaking, but that raw interestingness doesn't Do The Thing and resolve or "snap" to a relatable reference point. Which is really sad; without that extra "oh I get it!" this project is sadly going to have a hard time self-advertising and passively attracting interest. Even *cough* Mir (a Wayland alternative that's fallen into obscurity) has a tad more understandability going for it.
The upshot of this, though, is that if this ever gets to a point where it is relatable and can be appreciated at knee-jerk face value, it will be somewhere new and unfamiliar, and has the chance to have a noteworthy influence on how the Linux graphics status quo is reasoned about.
I think they just have strong design ideas and are rapidly blasting them out in stream-of-consciousness form. It’s a lot of information without as much structure as would sometimes be helpful, but it doesn’t seem like a TempleOS-type situation where they’re actually detached from consensus reality.
They made cool stuff but with a complete disregard for a lot of what the rest of the industry has been up to and as a result have and are trying to reinvent wheels they are under-qualified to reinvent. The display server is dope but they don't need to be trying to design a new paradigm of distributed compute in their networking layer, they got beat to that faster and better by other products.
The "rest of the industry"? The "Unix Desktop" industry? That one? The one that's still claiming that Wayland's deficiencies are due to it being a work in progress? The one that still can't synchronize on a single universally supported GUI library that's available by default on every system? He's not qualified to offer alternatives to _that_?
What sort of model glue should he begin sniffing in order to more effectively engage with this "industry"?
I get that some open source projects are focused on backend and infrastructure, but I wish they at least used a descent SSG with screenshots and what not to even figure out what something is.
wayland: if we move the compositor into the kernel we can solve several performance problems.
arcan: if we make a universal transport layer we can solve most topological problems.
Plus I like arcan because it reads like it is one guy, deciding the world needed a better display server. and just keeps at it, day to day, no team, no consensus, just getting it done.
however disclamer: I made a port of arcan to openbsd and could never quite figure out what to do with it. the curse of a display server with not enough software I guess.