Hacker News Reader

Comment by MayeulC | Hacker News Reader

I quite like suckless software, unfortunately it is very X-centric.

Not sure what their view on Wayland is?

I can't speak for the group, but my guess is in practice they would dislike the Wayland ecosystem. A cursory google seems to show this too.

A lot of their tooling and favored software works by being as minimal as possible while doing its one primary purpose well. They dislike monoliths like systemd, pulseaudio, networkmanager etc. which do many tasks. The wayland protocol itself is fairly minimalist, which they seem to like, but to practically use it you end up requiring a monolith compositor/window server.

Further, Wayland fundamentally doesn't support running e.g. in a non-compositing mode, and requires more cruft to draw to the screen vs just using xlib.

At the end of the day, all the functionality that Xorg provided has to be served by something. And when you're not using one of the big boy software stacks like Gnome or KDE, that means re-implementing the wheel, or copying from another project like wlroots. I suspect they prefer the status quo in that Xorg is at least proven and fairly battle hardened software.

Just idly guessing, though.

It kinda cuts both ways, I suspect. Xorg is itself a huge monolithic mess, but it takes all the messy bits into itself and lets you extend it with nice small programs (ex. window managers, screenshot programs, keybinding programs, whatever). Wayland is in some ways actually better because it removes the massive central mess that is the X server... but then forces every other component to deal with those things, and does so in ways that to date play badly with composability in the software (ex. now a "window manager" is a compositor and must do most of what the X server did before, and you can't factor out things like screen capture because that functionality is restricted).

> Wayland is in some ways actually better because it removes the massive central mess that is the X server... but then forces every other component to deal with those things, and does so in ways that to date play badly with composability in the software

This is a great example of Larry Wall's Conservation of Cruft principle. [0]

[0] https://metacpan.org/pod/distribution/Perl6-Doc/share/Apocal...

Exactly. It's why (I think) they haven't really spoken "publicly" about either directly, no listing on the "Software that rocks/sucks" pages, nor on cat-v or any related sites. Both approaches are pretty contradictory to their views on software. It's just that one is more composable than the other, and on top of that is older and a "known evil", I suppose.

cat-v.org is largely untouched since Uriel died in 2012; Wayland existed, but was enough of an experimental new thing in 2012 that I'd be very surprised if cat-v.org mentioned it.

That's fair. I guess I meant more that "network" of related sites and interests, of which I just kind of mentally group together as "cat-v", for better or worse.

The wlroots/Sway teams are leading the way towards a more Unix-y Wayland experience, though. Over the last two years or so, a lot of 1:1 replacements for those "do one thing and do it well" X11 tools have appeared: https://github.com/swaywm/sway/wiki/i3-Migration-Guide#commo...

> monoliths like systemd, pulseaudio, networkmanager etc

This is the first time I've seen networkmanager in that list. What's the alternative way of handling network connectivity? nmtui seemed pretty one-function to me

There's a lot of CLI tools in suckless too. I quite like their IRC client https://tools.suckless.org/ii/

Reply on news.ycombinator.com