All-In on Omarchy at 37signals
world.hey.com
world.hey.com
Getting used to hyprland and walker has been a short and eye-opening experience. I've had tiling before on Windows, but I've never been forced to use the tiling exclusively.
I'm all in on the vision behind Omarchy. I recognize DHH has a bit of an enthusiastic bent - He's been obsessive about Omarchy for the past month, and his opinions change. In some ways, he lives by the mantra "strong opinions, weakly held". I don't think that's a flaw.
I get the sense that this one is "the one" -- It's a foundation of linux that is entirely dependent on opinions, and DHH has them.
But, I think more importantly, the groundswell around the project highlights that there's a general dissatisfaction with the state of operating systems. The move to Linux has long been overdue for me personally. I'm incredibly tired of Windows and the Microsoft shenanigans. The adware on what should be a personal computer is an abomination.
I see Omarchy as an opinionated way of composing Linux in such a way that it offers a uniquely different premise of what an OS can be.
So far, I'm loving it, but am still tethered to Windows for some work stuff.
On the note of all the linux marketing, Jonathan Blow summed it up best:
> The people who would historically be excited about a new operating system can't do that any more, because everyone is too helpless to even conceive of a new OS.
> So they have to get excited about a mildly different arrangement of bloatware from That OS From 35 Years Ago.
> But as long as you give it a Cool Name, everything is good.
> Elon: makes car company (when everyone thinks electric cars will never work), rocket company (the rockets land themselves), Neuromancer brain chip company.
> Computer Nerds: Noooooo I can’t make an OS because drivers and adoption!!!!1
———
And from another thread
> It would be nice to have an OS with a proper job system as a core component. No legacy threads or mutexes at all. Everything is designed to be fine-grained parallel for modern 16+ core CPUs.
> For starters, every API is asynchronous command buffers with an optional slower/easier noob API on top. There are a lot of things that could tremendously simplify userspace as well.
I’m the furthest thing from an expert on this so I could be really missing important details, but isn’t this the idea behind Grand Central Dispatch from like 2009?
If I've learned any lessons in building software, it's that you very often have to strangle the old to make way for the new, but you get stuck in hell if you try to do it all at once. Getting Linux adoption up and general support improved, and then leveraging it as a WSL-esque bootstrap into a new OS might be a way.
On Komorebi, put simply, as an OSS maintainer I think me and that maintainer see the world a very different way. Would use it if not for the approach to licensing - does look cool!
That power bank is 1.2lbs, and the laptop is 2.9lbs. Carrying around 4.1lbs ain’t light.
Maybe I’m just getting old but I’ve long since moved past thinking it’s ok to be inconvenienced due to what I’d consider a product deficiency.
Note: I’m not hating on Framework. I’m assuming this is an inconvenience due to running Linux on the desktop.
With respect to laptops, there aren't enough people looking at every SKU, of every manufacturer, and then translating the possible power state combinations of each bus and link and adapter into the config tools preferred by this-or-that single distro. It would require be a massive, more-thankless-than-average undertaking. As a result, the median laptop is not optimized for battery life out of the box, even for distros that do try (to say nothing for those like Arch that intentionally eschew making choices for you).
With some tools that allow self-tuning, it's not difficult to get 8-10 hours of battery life on the newest x86 platforms (or more on ARM laptops). However, it isn't done for you. It is an area that desktop Linux needs to invest in improving on behalf of the majority of people, who are less inclined to experiment and tweak.
I have an MSI laptop with an 11th gen Intel chipset and 16inch screen. I run Debian Gnome and get an easy 6 hours of work and 4-5 hours of straight Netflix and nothing about the laptop screams power efficient...
I spun up Omarchy in a VM just to see what the fuss was about, and when I opened Neovim it booted into a plugin manager and started installing at least two dozen random plugins, including an extremely over-eager autocomplete config that filled my screen with snippet suggestions. I was instantly irritated.
The animations are nice, but it's not for everyone. I'm already productive with Sway, so all the little differences weren't worth the eyecandy.
I doubt there'd be much complaint about a dev who switched to Linux but chose Gnome over Hyprland. VS Code and other tools are going to work the same on either one.
Devs use some combination of the setup/suggested dx from their company, their own preferences/tools, and a large amount of informally socialized tools and setup that other devs at the company use.
Trying to shift deliberately towards a more collectivized experience can help a lot. When one dev goes to work with another, having a stable base to work from can help a lot. Source files tend to be in ~/code, we tend to use rvm for version management, we use Y LLM-tools with Z MCP servers.
Being able to dogfood your working environment together feels impossible & absurd, but leaving everyone out there doing whatever they cobble together feels at least as absurd.
37signals is also sort of home of Rails, where "omakase" and convention over configuration are deeply woven into the lore. The idea of them building something that just works out of the box & is tuned makes hella sense.
Beyond just cobbling together tools, having a practice can help so much. My god I've totally become such a more empowered vim user after switching to astrovim! Yeah the very amazing automatic tool config helps so much. Sure the modular lazy vim plugin experience and the slick as hell astrovim community packs have made layering in new tools & configuring existing ones mostly very easy. But holy heck, astrovim has a visual menu for the leader key. It turns vim from having an implicit depthful interface to an explicit visible interface.
Someone configured all these tools to show up somewhere, made a pattern for walking through tools. It's game changing: having folks make some sensible dev centric defaults and make them visible has made all the difference for me. https://docs.astronvim.com/#-features
> March 7, 2024: Committing to Windows
https://world.hey.com/dhh/committing-to-windows-2d6388fd
> June 6, 2024: Introducing Omakub (based on Ubuntu)
https://world.hey.com/dhh/introducing-omakub-354db366
Place your bets on what is next: nixOS? Haiku? OpenBSD?
> “About one-third of Basecamp employees accepted buyouts today” https://news.ycombinator.com/item?id=26998127
Looking back at January 2024, it seems like DHH started to reach his personal boiling point with that sort of nonsense, which seems to mark the start of the aggressive search for another suitable platform, out from under Apple's control.
https://world.hey.com/dhh/apple-rejects-the-hey-calendar-fro...
https://world.hey.com/dhh/apple-s-new-extortion-regime-to-ke...
While the original pursuit may have been less about seeking Windows or Linux than it was about getting away from Apple, it does seem like the end destination in Linux has been pretty satisfying. And it's great to see someone investing time and energy into promoting a cohesive linux desktop user experience that isn't just focused on feature parity to something else.
As a developer, I am much more productive with Linux than any other OS. The workspaces feature is better (on most distros I tried), the commands more closely align with the prod environment, no need for complex virtual machine setups, no weird delays or missing critical commands when using bash. Everything just works and performs.
IMO, this is the payoff of having architected the system properly from the ground up, it's more stable. You don't get so many weird glitches as you do on Windows.
Windows NT's architecture is perfectly fine and in some respects superior to Unix, but Microsoft has made subsequent compromises that reduced stability.
I checked my git-log from my dotfiles. It says I installed hyprland 2023-07-23.
> 319649c 2023-07-23 (sbinnee) install(hyprland): wayland wm tag: hyprland
I used to run i3, dwm, bspwm (my favorite on X11) and tinkered with other wms. Since this commit, I have been full-time on Wayland. When Hyprland were young, it certainly had rough edges. But these days I don't feel overwhelmed even when I update several releases at once. It is stable and just works. The creator of hyprland, vaxry, is an incredible developer and maintainer. He's made so much progress on usability of Wayland.
After many years using Ubuntu, I migrated to Omarchy this weekend (Arch Linux + Hyperlnd, a tiling window-manager). Looking great so far!
I think a lot of people are sick of supporting poopy companies.
I'm sure it is very configurable, but every visual I've seen of this thing looks awful and not something I'd want to look at while working. But I understand we all have different tastes.
But even in the blog post I'm struggling with 'why?' here. Am I to understand the primary benefits here are improved battery life and increased developer productivity by tests running faster? Is that it?
I travel an inordinate amount and have never found a Macbook's battery life to be insufficient. I struggle to even remember the last time I've used my computer long enough to drain the batter and not be near a power outlet. I work from ski lodges, planes, my car. This has never been a problem for me. Not once. This just feels like a really bad metric to optimize for given a typical developers' schedule and work arrangement.
> On the flip side, we'll get a massive boost in productivity from being able to run our Ruby on Rails test suites locally much faster.
Is this not just a Ruby issue? I don't know what's basecamp or HEYs codebase looks like on the inside, but they don't feel like projects whose tests suites should require a completely different OS or hardware arrangement. I haven't used Ruby in a decade but I do recall it being frustratingly slow. This seemed to be an understood and accepted reality amongst teams that adopt it.
Anyway, I feel like a better 'why you should do this' in order, especially if it is being mandated amongst developers in a company.
Pretty sure it's a Docker on Mac issue.
I personally use all three major OS, Mac at work, Windows for gaming, Linux for everything else. For me different OS serves different purposes and gives me task-oriented productivity boost.
I respect that and truly admire what dhh has done and tweeted about over the last year or so. But for a business to mandate an operating system for their developers (which I fully agree they have the right to do) with those reasons is underwhelming.
What is a productivity boost for you in one OS might be a burden to someone else. And vice versa. That's the benefit of letting developers choose their OS.
I've been using for more than a decade and have never found it to be "frustratingly slow", whatever that means in practice. Except when running on Docker on a co-worker's Apple product, which I do not use.
Anyway, obviously Ruby (or any other dynamic language for that matter) isn't the solution to every problem. I was once tasked to refurbish an old Delphi desktop application that did some thermal simulations; I wrote the interface as a Rails webapp and the simulation itself as a compiled C routine, which I called from Rails via FFI.
[1] https://www.youtube.com/watch?v=mzDi8u3WMj0 [2] https://world.hey.com/dhh/living-with-linux-and-android-afte...
Lots of developers are tired of being hemmed in and disrespected by Apple. Omarchy gets us back into using an OS made for developers, by developers.
you certainly can. Omarchy doesn't appear to be anything special, just a tiling WM (of which there are plenty of on macOS: aerospace, yabai, amethyst, etc.) with some preinstalled applications + basic dotfiles. people have been running similar setups for years on macOS
I like parts of omakub, but impossible to use alacritty on Windows. Paste is broken. Known WONTFIX.
I would use a tiling window manager, but not ready to give up on the mouse. Title bars are sometimes nice.
Lazyvim not for experienced vi users.
I'm not anti-libertarian, I share some of the values, and I don't even disagree with some of the technical choices at 37Signals. However I think there's a lot of hypocrisy in the space, and DHH frequently comes across as being very tone-deaf and unaware of the impact his decisions have on others, basically an extreme lack of empathy.
This very well describes many things I saw working in big tech however. Perhaps fully owning an internal implementation adds more weight to the decision than purely quality of the product itself.
In reality, this change primarily applies to the engineering team and will be phased in over several years. The company as a whole will continue to run Windows, macOS and Linux.
There's also a get-out clause for any developers who just can't get along with Linux and prefer to stick with their beloved Mac.
Source: the podcast episode linked from the DHH post (https://37signals.com/podcast/moving-to-omarchy/).
This seems wildly subjective. Can you give some examples of how Omarchy is worse?
I find using macOS to be generally frustrating and cumbersome to use and I'm regularly confused when people say it's unequivocally the better choice for all things.
It's very easy to quickly build something that solves 80% of your problem, and because it's tailored for what you need and you understand it, it's easy to believe that's better than the existing tools. The problem comes in the long term, in supporting more use-cases. You can decide not to support more, but that's a sacrifice (and in the case of Omarchy the end users may bear the brunt of that), or you need to put the time in to develop it.
Omarchy will have less development effort than macOS, their deployment system will have less development than Docker/Terraform/etc, and that's all a trade-off.
But… their deployement IS docker and terraform and etc. Their platform is Linux, nvim, etc. In total there is MORE development than all of apples platforms. It’s just not unified under one umbrella.
Anyways you have a false equivalency there. Something being "less developed" doesn't automatically make it "worse". On the contrary, projects with larger active development effort generally trend to enshittification whereas smaller projects can stay focused on solving their core use cases.
For a macOS-specific example, see https://x.com/ryolu_/status/1954598993515962707