Linkblog/2025/04/07

Chicken Jockey.

I saw the movie tonight:

Honestly, as far as video game adaptations go… I think it actually did a pretty good job, it was freaking corny, but in a self-referential aware way.

Spotting DanTDM, the Technoblade reference, and Jens, was also fun.

An unfinished version of the movie has leaked online as well, I might have a copy of it on my computer…

Its really funny.

I have met a lot of developers in my life. Lately, I asked myself: “What does it take to be one of the best? What do they all have in common?” […]

Matthias has a lot of great points here, I myself attempt to put them into play (sometimes).

Read the Reference

If there was one thing that I should have done as a young programmer, it would have been to read the reference of the thing I was using. I.e. read the Apache Webserver Documentation, the Python Standard Library, or the TOML spec.

This 100 times, the amount of times I see beginners reaching for W3Schools over MDN…

Break Down Problems

Everyone gets stuck at times. The best know how to get unstuck. They simplify problems until they become digestible. That’s a hard skill to learn and requires a ton of experience. […]

Don’t Be Afraid To Get Your Hands Dirty

The best devs I know read a lot of code and they are not afraid to touch it. They never say “that’s not for me” or “I can’t help you here.” Instead, they just start and learn. Code is just code. […]

Never Stop Learning

Some of the best devs I know are 60+ years old. They can run circles around me. Part of the reason is that they keep learning. […]

Both Docker Desktop and the CLI has AI in it now.

Why?

Because everything needs that now for some reason.

New Jane Remover is good.

The second track goes hard:

Developer: Wise Compiler Crab, I seek to unravel the mysteries of my u128 sword. How can it wield such power on a realm ruled by 64-bit registers?

With with barbed remarks, Graham explains how u128 actually works even though the plan fell off.

To ensure Next.js can be deployed anywhere, including severless platforms with custom requirements, we are planning to add deployment adapters. Vercel will use the same adapter API as every other partner.

I wonder if this will mean OpenNext won’t be needed anymore, I would assume the OpenNext folks would embrace that maybe, as maybe this is actually an answer from Vercel of how to run it serverless just like they do.

I spoke about Edurado’s article on Next.js the other day, and this was one of his big gripes as someone working on Next.js support at Netlify.

They even reference Netlify directly:

However, it has been difficult for serverless platforms like Netlify to support deploying Next.js due to their custom requirements. After talking with their team and others, we want to make Next.js easier to maintain for deployment platforms.

Hopefully Edurado’s gripes get resolved, but this is at the RFC stage, I feel like most won’t trust them before action is taken to actually support this stuff, not just talk about supporting it.

When programming in Odin you can use arena allocators. If you use an arena allocator combined with a dynamic array, then there are a couple of pitfalls that may not be apparent at first. Let’s look at what arenas are, how you can run into trouble when naively using them with dynamic arrays and what you can do instead.

As always, Karl is making dope Odin content again.

Karl also has a video companion to this article as well:

Folks what that haunt me (positive) on the Fediverse may have seen me sharing progress shots from this, but here I am, and I have made another camera application for the web. Leibovitz combines a lot that I learned making my other camera applications into one, hopefully less clunky package.

Put some images you have into the tool and play around with it, its quite fun.

Look at Peanut, he’s so amazing.

Building a native GUI for a cross-platform application is a decades old problem. Nowadays, most people just don’t and fallback to using a non-native experience such as Electron instead.

One approach to building a native GUI for a cross-platform application is to write all of the business logic in a cross-platform language (C, Rust, Zig, etc.) and then write the platform-specific GUI code. This is the approach I take with my my terminal emulator and it works really well. As of the current date writing this post, 93% of my repository is business logic in Zig and C, and 4% is macOS-specific GUI code in Swift.

I’ve been using Ghostty for a while now.

I managed to get into the beta a year before release, it was already to the point I had no bugs with it so to me it was simply a dope terminal emulator I felt like I had clout for having access to.

The design of the application to me has been interesting, I’ve used it on both Linux and Mac, so I’ve used SwiftUI Ghostty, and GTK based Ghostty: both feel native to their platforms.

While I think Electron / Tauri / Qt / etc. are good choices for most, I do think if you have the time and skill, and bother to learn the big 3 (or be like Mitchell and just ditch Windows support so its just the big two sans the big one to you), why not?.

I’ve got a project I think I might actually use this structure for.

I’m thinking Odin for the core of mine, because I’m an Odin enjoyer, SwiftUI for Mac (duh), GTK for Linux, WinForms for Windows.

I think targeting a OpenGL version that works on all 3 would be good for the first version, maybe having a Metal native version like Ghostty has for Mac would be cool, but maybe not worth the extra work.

Thinking Time Exercise

Sit somewhere quiet

Hit the bathroom, get some water, etc. first so your body doesn’t find an excuse for you to get interrupted.

Write a high-quality question at the top of a blank piece of paper

Don’t use a computer or phone. The author uses the same “thinking chair” and has a dedicated journal.

Set a timer for 45-60 minutes

Best to use a dedicated timer instead of a phone, to avoid getting distracted and to lean into “analog mode”.

Write and think about this one problem until the timer finishes

Do this without interruptions - no querying, or web or phone use. The goal is write a lot, without editing. Your goal is ideas and possibilities, not absolute answers.

When you’re done, read it over

This might give you answers, or may give you direction for another thinking time session. You may find that your best works shows up later in the session, so make sure to sit through the whole 45-60 minutes (I’ve found the best ideas come in this extra time in group working sessions also).

This sounds good, I’m not sure what sort of questions I’d personally write down on a piece of paper for this sort of session, but I’d honestly like to give it a go.

Software generated entirely by AI is not an hypothetical future — it’s a reality today. I don’t think it’s an overstatement to say that this is the largest shift in the history of this industry, unravelling before our eyes at an astonishing pace.

[…]

I take joy from different aspects of writing code on a daily basis: writing some tests and seeing them pass for the first time, dissecting the root cause of a gnarly bug, or coming up with an elegant solution for a seemingly impossible problem (and always patting myself on the back for it).

The realisation that computers can now perform these tasks in a fraction of the time it takes me is unsettling. It’s impossible not to ask: will AI replace me?

[…]

As I let those feelings simmer, I wanted to take a step back and look at software as more than just something that I enjoy creating; it can also be a force for good with a profound impact on people’s lives.

Coming to the terms with the state and direction of the software development industry has not been easy, I both agree with and relate with what Eduardo is saying here, it sucks that the art of unassisted programming might become more and more a thing required for only the most bespoke of code, or just for fun.

This means that the amazing projects we see today don’t represent all the creators, the ideas, or the potential out there — they’re just the ones that made it through the creator’s funnel. A classic case of survivorship bias.

There’s some discussion on Twitter regarding that Quake II AI demo mentioned yesterday, with John Carmack himself chiming in calling it “impressive research work!”.

This is honestly close to my current interests, as I’m looking into taking Quake II and turning it into something transformative / fun, and doing it by the seemingly “old fashioned” programming approach.

So yeah, my reaction to the thing yesterday, was simply, “Ew!”.

But, after considering Edurado’s post, I’m less jaded about it, I do think its not super useful to gatekeep the ability to make games, it is an endeavour to learn all the component parts required to make a game.

But also… a one-shot prompt to produce something…

I don’t really feel like that is creation still, similar to the current space of generative imagery, purely generative games don’t rub me the right way.

Carmack did mention impressive research work, and I think generative games are also interesting research work, but I also know as soon as they are somewhat good, they will be packaged up into a consumer product and could even negatively affect the indie space.

I’m conflicted.

Anyways.

New Wes Anderson movie is a Wes Anderson movie.

As always, great trailer, so many faces, I will be seeing this in theaters.

Tobi posted an internal memo at Shopify that he assumed would leak, so he himself posted it to Twitter.

I feel like I agree with aspects of this, with regards to integrating AI into workflows, having it codified at the employee review level seems… interesting, but I think for the type of company Shopify is, I get it.

This point didn’t feel great though:

Before asking for more Headcount and resources, teams must demonstrate why they cannot get what they want done using AI. What would this area look like if autonomous AI agents were already part of the team? This question can lead to really fun discussions and projects.

I get what this is saying, but this reads to me like: we’re going to hire less people to do software because AI can do the work of more people.

Just feels like the productivity curve of an individual contributor will be going up with a smaller workforce.

I do think being transparent about this is better than having an internal memo that leaks though, but I assume Tobi is the type to do this to mostly boast to other CEOs in the space than actually start a conversation about anything.

The rsync command line tool has long been included on macOS, but Apple has provided the last version of rsync 2.x (rsync 2.6.9, released in November 2006) and did not update rsync past that even though rsync 3.x was released.

[…]

Now with macOS Sequoia, Apple has replaced rsync 2.6.9 with openrsync, an implementation of rsync which is not using any version of the GPL open source license. Instead, openrsync is licensed under the BSD family of licenses, specifically the ISC license.

Intersting, MacOS rsync is now using OpenBSD’s openrsync.