This is where I write about the technical side of things: tutorials, concept explorations, and the occasional personal reflection (including those yearly "what just happened" posts).
If you're looking for the more strategic stuff dealing with leadership, architecture decisions, the fractional CTO angle, that lives on my Substack.
Picture this: a designer spends two weeks crafting a perfect card component. Tight heading, one line of supporting copy, a striking image. The proportions are deliberate. The whitespace is intentional. It's a small, considered piece of art.
There's something counterintuitive about AI coding tools and technical debt. The promise is that they speed things up. And they do! But speed without direction is how debt compounds, not how it gets paid down.
Without guardrails, AI assistants learn context from the code around them. Which means they inherit the debt too. A deprecated API that appears in enough files gets used again. An inconsistent pattern gets replicated with confidence, at the speed of autocomplete. It doesn't introduce new debt out of nowhere. It faithfully copies and amplifies the old debt, everywhere, consistently.
There's something counterintuitive about AI coding tools and technical debt. The promise is that they speed things up. And they do! But speed without direction is how debt compounds, not how it gets paid down.
Without guardrails, AI assistants learn context from the code around them. Which means they inherit the debt too. A deprecated API that appears in enough files gets used again. An inconsistent pattern gets replicated with confidence, at the speed of autocomplete. It doesn't introduce new debt out of nowhere. It faithfully copies and amplifies the old debt, everywhere, consistently.
There's a pattern that keeps showing up in software engineering: the tools that give us the most leverage also demand the most structure.
Version control unlocked collaboration at scale, but only after teams agreed on branching strategies. Linting and formatting took out a lot of opinions during code reviews, but first we needed to have settled the "tabs vs. spaces" debate. CI/CD pipelines sped everything up, but only once contributors agreed upon a definition of done. AI coding tools are following the same arc, and the teams getting the most out of them have already figured this out.
If you're (like me) liking the idea of having an AI coding assistant that runs completely locally on your machine (privacy!), there's a good chance you've stumbled upon Goose-CLI. Combine it with Ollama for running AI models locally, and you've got yourself a pretty powerful setup that doesn't send your code to the cloud.
Let's see what we need to do in order to get both systems up and running, configure them to work together, and set up some convenient shortcuts to switch between coding and writing tasks.
If you're (like me) liking the idea of having an AI coding assistant that runs completely locally on your machine (privacy!), there's a good chance you've stumbled upon Goose-CLI. Combine it with Ollama for running AI models locally, and you've got yourself a pretty powerful setup that doesn't send your code to the cloud.
Let's see what we need to do in order to get both systems up and running, configure them to work together, and set up some convenient shortcuts to switch between coding and writing tasks.
By the time you're reading this, the plane that will land our family on Bonaire will have touched down on the tarmac at Flamingo Airport.
This year felt a bit quieter on the outside, but when I look back at everything that happened it was anything but dull. A conscious decision to chase fewer big highs and lows. More small moments that added up to something meaningful. (And we'll get to that in the end!)
TL;DR: I upgraded a 32k LOC Nuxt app from v3 to v4 in 29 minutes of actual work. Here's how it went and what I learned.
I'm maintaining a reasonably sized Nuxt application - not massive, but substantial enough to make upgrade decisions matter. We're talking about a project (excluded tests here) with 32k lines of code spread across 173 files, with a healthy mix of 72 Vue components and 44 TypeScript files. The dependency footprint includes 34 production dependencies plus 7 development ones. So, not that big, which makes it a perfect candidate for an upgrade scenario!
TL;DR: I upgraded a 32k LOC Nuxt app from v3 to v4 in 29 minutes of actual work. Here's how it went and what I learned.
I'm maintaining a reasonably sized Nuxt application - not massive, but substantial enough to make upgrade decisions matter. We're talking about a project (excluded tests here) with 32k lines of code spread across 173 files, with a healthy mix of 72 Vue components and 44 TypeScript files. The dependency footprint includes 34 production dependencies plus 7 development ones. So, not that big, which makes it a perfect candidate for an upgrade scenario!
If you're (like me) liking the ease of hosting with Netlify and are using Nuxt to power a static site, there is probably a rule of thumb which predicts when you'll need a form on your site. Luckily, Netlify has support built in, but you need to set up your configuration correctly for it to work.
Let's see what we need to do in order to add a concise form (you'll know how to take it from there, I'm sure) and get submissions in your Netlify dashboard.
I recently completed a rite of passage: I've published my first app in the Google Play app store. The reason for going through this was two fold: I was aware of the capacities of multiplatform builds using my favorite framework of choice and wanted to explore how this would pan out in practice. At the same time, the perfect opportunity presented itself: my kids were practicing tables of multiplication and to me, that seemed a perfect challenge to build an app for!
I recently completed a rite of passage: I've published my first app in the Google Play app store. The reason for going through this was two fold: I was aware of the capacities of multiplatform builds using my favorite framework of choice and wanted to explore how this would pan out in practice. At the same time, the perfect opportunity presented itself: my kids were practicing tables of multiplication and to me, that seemed a perfect challenge to build an app for!
Ah, the JavaScript Date object. It just doesn't cut it, does it? Working with dates can be daunting or maddening! So rather than using the Date object, we've been resorting to libraries (like moment.js or date-fns) that tackle the complexity for us. We're getting at a comfortable space right now, with the Temporal API!
It's not widely supported right now (experimental stage), but close to being finalized and adopted by major browsers!
When working on pull requests on github, use the comment system or in text documents in your repository, the text is converted to and displayed as markdown. Here's a couple of tricks you can use to leverage some features for better readability and clear communications when cocreating in github.
With the second major version already released, there's a buzz surrounding Astro and it's hyper performant content focussed web applications. Apart from the focus on speed, Astro is also labeled itself as a "all-in-one" web framework.
All-in-one meaning that it supports most modern frontend frameworks with zero configuration. I decided to research that claim and I am pleasantly surprised!
With the second major version already released, there's a buzz surrounding Astro and it's hyper performant content focussed web applications. Apart from the focus on speed, Astro is also labeled itself as a "all-in-one" web framework.
All-in-one meaning that it supports most modern frontend frameworks with zero configuration. I decided to research that claim and I am pleasantly surprised!
For this very blog I sometimes want to do internal linking. A common practice for my writing is to have a heading that marks the beginning of a relevant section. Rather than managing the anchor links by hand, I used the `documentToHtmlString` method (provided by Contentful to convert the document structure to HTML structure) to do it for me and generate predictable anchors.
With AI systems on the rise, I think it's valid to take a look at where they can play a role in supporting our tasks now and for the future. The OpenAI platform allows you to make use of published algorithms in a fairly straightforward way. It's the perfect gateway drug into AI and Machine Learning!
In this tutorial we'll take a look at a specific API, namely the one that helps you in intercepting user input by applying natural language interpretation to use as moderation.
With AI systems on the rise, I think it's valid to take a look at where they can play a role in supporting our tasks now and for the future. The OpenAI platform allows you to make use of published algorithms in a fairly straightforward way. It's the perfect gateway drug into AI and Machine Learning!
In this tutorial we'll take a look at a specific API, namely the one that helps you in intercepting user input by applying natural language interpretation to use as moderation.
If you host your application or website on Netlify and have it intregrated with Github, you have the advantage of pretty good automation in your pipeline on opening a merge request.
If you also happen to have end to end (e2e) tests, you can easily point them to the preview deployment, to automatically verify that your vital features are still functioning!
Now that you have a way of generating a static website, you have the opportunity to host it on a cheap low cost (usually free) tier of a wide variety of hosts, such as Netlify, Vercel, Github Pages or basically anything that can serve an HTML page.
Now that you have a way of generating a static website, you have the opportunity to host it on a cheap low cost (usually free) tier of a wide variety of hosts, such as Netlify, Vercel, Github Pages or basically anything that can serve an HTML page.