TL;DR: you can check how many dogs I’ve seen during my runs here.

I mentioned in a week update that I started counting the dogs I see during my runs. I realized that it was a good way to keep my mind just busy enough to enjoy the run without overthinking everything I have going on in my life right now, the news and so on. At first, I put those numbers in an Obsidian note, but then I realized it would be fun to make a tiny website for it.

The website uses 11ty. I used it to create the website for Sound Games and I really like it, even if the documentation feels a little “draw the rest of the owl”-y at times.

I’m storing the data in a basic JSON file:

[
    {
        "date": "31/12/2025",

        "location": "New Westminster",

        "distance_in_km": 9.47,

        "count": 72
    }
]

Additionally, I’m using a data file named eleventyComputed.js that the engine will load if it exists, which allows you to transform existing data. In my case, I’m using it to compute the total number of dogs and store it into a variable named total that’s available in my layout:

module.exports = {
  total: (data) => data.stats.reduce((acc, stat) => acc + stat.count, 0),
};

Finally, since it’s a very simple website to host, I decided to experiment a little. I followed this article and decided to put it on Bunny.net (shameless referral link), in the hope of moving this website out of Netlify. It didn’t take me long, I just had to create a storage zone, create a pull zone, configure the DNS and upload my generated website. I didn’t do the continuous deployment dance yet, because even I knows that would be overkill at this stage.

It was a really fun project to kickstart after spinning it in my head for a couple of days. I have a bunch of ideas that I want to implement because it’s a little boring-looking right now and it’s really hard to ignore that I lazily used Bootstrap. A project like this needs to be at least 46% more whimsical. In the meantime, you can check how many dogs I’ve seen during my runs here or check out the source code here.

I paused my lessons with the new trainer I’d found for running, because after taking a step back and thinking about it, I realized it wasn’t the right approach for me. Where I should have felt empowered and motivated, I realized I felt like shit when I wasn’t able to run one day because of… well, life. Instead I switched back to my original plan, which is meeting regularly with a physio I like who happens to specialize in running. We’re going to plan for some strength training on top of my homework runs to stay injury-free until the big day. I’ve also aquired a static bike for cheap on Facebook Marketplace, just because. As a result, my socials’ feed are now full of ads for a 2000$ peloton bike.

Speaking of homework runs, the shop that organizes the running club I’ve been a part of for a couple of years now, has opened a new location. The original shop is in Burnaby, the new one is in Kitsilano. To celebrate the event, they organized a 19km run between the two locations. I struggled a lot on the second half, because of poor fuel and water planning on my part, but I’m glad I did it. I’m even more glad they’d hired a bus to take us back after. I only counted 48 dogs though, which is a little disappointing for a 2hours run.

Finally, I’ve spent a few hours printing a tool to re-spool some filament. It was probably my fault, but a reusable spool of filament sprung open during a print, causing the filament to run amok when I wanted to unload it. As a result, I needed to find a way to re-spool the filament, which meant printing a device dedicated to this. There is a running joke in the 3D printing community that once you get a printer, you mostly end up printing stuff for the printer itself… I guess it’s true in this situation. I haven’t attempted to respool yet, but I’m sure it will be a fun update next week.

Links from my queue

• Starting with a weird one, did you know that wombats poop cubes? I fell into a rabbit hole after someone pitched a game called “shitting bricks”. Oh, they also fight using their ass.
• Those two posts about macOS Tahoe’s choice of icons and the struggles with resizing have been really painful to read. I think I’ll be sticking with Sequoia for a little while.
• I found this article about ditching Bear quite interesting, at a time where I’ve found myself taking way more notes than before. It’s spread across my physical notebook, my deskbuddy and of course Obsidian. It did cause me to do my regular tidying of notes a little earlier than usual, though.

I’ll end things with a shameless plug for work-related stuff, but if you’re into chaotic (yet fun) games that you can play with friends, please add Go, Ape Ship! to your wishlist. If you’re curious about what we’ve been doing at Sound Games, follow us on Mastodon here.

This past week as been composed of resting and chilling with my partner. We started an overdue rewatch of Schitt’s Creek, finally watched the first episode of Pluribus and cooked a lot of good food. The past few years have not ended well for us. It became almost a sad tradition to have to travel back to France at the last minute because of a family emergency of some kind. This year was quiet which we really appreciated.

This year I’ve signed up to run the BMO marathon in Vancouver. A friend planted the seed in my head last year and I toyed with the idea for a few months before finally signing up. I’ve been running for long enough that I know I can run a decent half-marathon without training too extensively, so I wanted to challenge myself and see if I could commit to a more demanding schedule. I didn’t set any goals besides training properly and hopefully finishing the race. During the past couple of weeks, I’ve been slowly ramping up, running longer distances and counting the dogs I see. I’m supposed to run 19km on Saturday, so that should be interesting. I’m excited to meet weekly with my running buddies every Saturday morning for the next 5 months.

I finally took the time to look into the issue preventing me from using Bambu Studio 2 and I’ve been printing a lot of different objects, from stackable boxes to sort LEGO or puzzle pieces, to a lithophane made for a friend. I took a fun quote from their child and made it into a tiny box that lights up, which was pretty fun.

I was fairly excited to go back to work today. We’ve released Spilled! and announced Go Ape Ship! last year (as you can see here), so things have been moving fast.

I’m planning on sharing links of stuff I’ve enjoyed in the past week but I didn’t prepare them ahead of time and it’s already Monday. For this week, I just want to give a shoutout to this transcript of a talk about the CI infrastructure for Servo. I was a little thrown off by the format at first but I quickly got into it and couldn’t put it down. Truly fascinating stuff.

My printer is a Bambulab P1S that I really like. As my friend told me before buying, it’s pretty much like if Apple had made a 3D printer. I don’t know how much this statement holds true these days, considering the drop in software quality, but I must admit that I’ve had very little trouble with my printer ever since I plugged it in.

I don’t remember when I tried to update Bambu Studio (the app you use to talk to the printer) to version 2.x but I remember running into an issue when I tried to start a print, something about a nozzle issue. I didn’t really want to look into the issue immediately, so I’d just reverted back to the latest 1.x version and moved on with my life. Based on the release notes, I suspect that was around February 2025. Since then, I’ve kept printing with version 1 of the software, ignoring the occasional warning telling me that the 3D model I’d downloaded would work better with the latest version of the Bambu Studio app.

Today, as my lovely partner asked me to print some cute Octopus Chip Clips, I decided to download the latest version of Bambu Studio again. As expected, I ran into an issue about a nozzle.

🚨 Invalid nozzle information, please refresh or manually set nozzle information

I felt many emotions at once: confusion (What do you mean the information about the nozzle is invalid?), resentment (It’s not my fault! I haven’t changed anything!), and finally anger (it’s your hardware! I thought you were the Apple of 3D printers?!). Once I’d stopped crying and screaming under the rain (of which we’re currently getting plenty #vancouver), I decided to start looking around for solutions.

After a little while, I decided to make sure that the printer was running the latest version of the firmware. It would be a little emnbarassing for me to be running outdated somewhere, as someone who keeps writing unoriginal blogposts about keeping your dependencies up to date. I selected the “Device” section of Bambu Studio, clicked on “Update” to see that my printer was running the latest version of the firmware, 1.04.something. Cool.

Still somewhat unsatisfied by this answer, I went looking for a list of the firmware’s versions to double check where it said the latest version was 01.09.00.00 and that version 01.04.02.00 was released… in October 2023. My printer has been running a version of the firmware released 2 years ago. Embarassing. Fortunately, I took a quick peek at the release notes and didn’t see any release made to prevent the printer from catching on fire.

The question is, why did Bambu Studio think the printer was running the latest version of the firmware if it wasn’t? The answer is pretty frustating. You see, there are two ways you can run your printer. You can either connect to Bambu Labs’ cloud services or you can run your printer in LAN only mode. While I wasn’t super worried about seeing my printer going rogue after some kind of targeted attack against Bambulabs’ cloud services, I didn’t see the value of connecting to their cloud services either.

Bambu Studio is open source, so it was easy to locate this open issue. The short version is that if you’re running in LAN only mode, Bambu Studio will always think you’re running the latest version of the firmware. At least, that was the case until September 2025 but to be able to get local only updates, it looks like you need to connect your printer to the cloud at least once.

I did that and the printer immediately asked me if I wanted to update to the latest version of the firmware. I did the update, the nozzle issue went away like magic and I was able to print some very cute chip clips.

These days, I use Kotlin for server side development. I already covered here why I like it and not much has changed since. To go a little deeper into our technology stack, we use Postgresql (with the exposed framework) and describe API specifications with OpenAPI (using Fabrikt for code generation). I’m mentioning those because I had to update them all at once in a massive pull-request impacting around 150 files.

Earlier this week, I ran my script that checks for outdated dependencies in a gradle project and realized there was a bunch of libraries I wanted to update, including the kotlin plugin itself, kotlinx-datetime and exposed.

Updating the version of kotlinx-datetime introduced deprecation warnings: there is no longer need for their Instant and Clock types, since those are now available in Kotlin’s standard library (since version 2.1). Turns out, representing time is a common need, so it made sense to move it there. Unfortunately, updating kotlinx-datetime broke Exposed’s date and time types.

As a result, I switched to a different branch and updated Exposed to 1.0.0-RC-4, which was… a lot of work, but pretty straightforward package changes (org.jetbrains.exposed.sql.Table is now org.jetbrains.exposed.v1.core.Table, etc). It worked until I realized that Exposed depends on… a more recent version of kotlinx-datetime.

After a quick search it looked like I could have used a compatibility version of datetime, but it seemed easier to go for it, merge both branches together and address the errors that popped up.

Lastly, the last task was to upgrade Fabrikt, because the generated code expected to use kotlinx.datetime.Instant and not the never kotlin.time.Instant. After bumping to the latest version and tweaking the code generation command to set the --instant-library option to KOTLIN_TIME_INSTANT, we were back in business.

A pull request that touches around 150 files is way about my comfort threshold, but an extensive suite of tests that covers all the sensitives parts made it a lot more tolerable.

TL;DR: Even if the documentation states that PlayerPrefs stores player preferences in a plist file on macOS, this is partially true since it uses the defaults system. If you’re trying to make changes to the state of your game, use defaults write com.company.game key value from your terminal.

For context, we’re working on a game where we store state (last time you did a specific action, the preferred volume for the music, inverting the Y axis on your gamepad like some kind of monster…). Right now we don’t have a “reset to defaults” button which means I couldn’t re-test actions that happen the first time you launch the game, for example. It’s in the work, but apparently even in the world of video games people try to assign priorities to different tasks and prefer working on the more pressing ones first. Weird.

As a result, I started peeking into the source code of the game, harassed my colleagues on the other side of the Atlantic to figure out what we were using and learned about PlayerPrefs which, as the name suggests, stores player’s preferences. Further, as the documentation says, the actual preferences live in a flat file in ~/Library/Preferences/ExampleBundleIdentifier.plist where ExampleBundleIdentifier is your bundle identifier (duh) such as com.mycompany.mysupergame. This was enough to get me going. I grabbed my old and weird friend PlistBuddy that allows you to decode and execute command against a property list file: print content, edit a key, delete it… I should probably have moved to the more modern version, plutil, but that is not here nor there. Also, I forgot it existed until I read the wikipedia page linked earlier.

I ran some PlistBuddy -c "Delete :my-state-key" codes.romain.whatever.plist commands, relaunched the game and… nothing changed. Even worse, the changes I’d made to the file had somehow never happened. After some more digging, asking around and trying to search online for an answer, I brought up the issue with my colleagues and someone mentioned iCloud potentially being the culprit. That’s when it clicked, like a movie where the hero (in this case, me) has a flashback and remembers something from their childhood (in this case, the last time I worked on a macOS app) that will save the day: user defaults.

If you’re not familiar, when making apps for the Apple platforms, there is this simple key value database you can use to store informations within your app. I won’t go into details, but you can access this database on macOS using the defaults command line tool. On a hunch, I typed defaults read com.mycompany.mysupergame and saw the informations I was expecting to tweak, and more. I tried to update some value using defaults write ..., relaunched my game and like a Christmas miracle, the changes were showing up in the game.

What’s the moral of the story? I can think of a couple. First, if you have to use fs_events to watch the changes happening to a file because you’re convinced that a goblin living somewhere in the operating system is modifying that file in secret, you probably need to take a break because you’ve gone too deep. Second, always remember to ask your colleagues, because someone will suggest something that will put you on the right tracks. Lastly, because you’ve only been through the first part of a 27 hours long unity tutorial doesn’t mean you can’t make meaningful contributions to the world of video games. Believe in your dreams.

Keeping my dependencies up to date is something I keep on writing about, most recently to introduce a command line tool I wrote to list out of date dependencies in Swift projects. Kotlin has replaced Swift for me these days, but the problem remained, especially as we’re building cool stuff at Sound Games and I get to make the servers go brrrrr¹.

Dependabot is one of the most popular ways to keep your dependencies up to date. It started as a standalone product and is now part of GitHub’s offering. As you may know, I’ve decided to stop using GitHub, but that’s not (only) why I’m not using Dependabot on other projects. I think that updating dependencies should be a conscious decision. Just because you get a pull-request created automatically where all your tests are passing does not mean that the update is safe. It’s actually pretty easy to bite yourself in the butt by merging a green PR, going to bed and waking up to angry message from the 12 daily users of your side project about comic books².

Now for the perfect solution: I don’t really have one. Recently at work, I created a super quick and dirty script that retrieves all the top-level dependencies we use in the project, retrieve the most recent versions available on the maven central repository and print a sorted list to compare both. I figured I would write about this.

Listing dependencies

You can list your dependencies in a Gradle project using the gradle command line tool to run the :dependencies subtask. It will print a tree with all the dependencies used for a specific configuration. Something that roughly looks like this:

|    +--- io.ktor:ktor-utils-jvm:3.2.3 (c)
|    +--- io.ktor:ktor-serialization-kotlinx-jvm:3.2.3 (c)
|    +--- io.ktor:ktor-call-id-jvm:3.2.3 (c)
|    +--- io.ktor:ktor-network-jvm:3.2.3 (c)
|    \--- io.ktor:ktor-io-jvm:3.2.3 (c)
+--- io.ktor:ktor-client-core:3.2.3
|    \--- io.ktor:ktor-client-core-jvm:3.2.3
|         +--- org.slf4j:slf4j-api:2.0.17
|         +--- org.jetbrains.kotlinx:kotlinx-coroutines-core:1.10.2
|         |    \--- org.jetbrains.kotlinx:kotlinx-coroutines-core-jvm:1.10.2
|         |         +--- org.jetbrains:annotations:23.0.0
|         |         +--- org.jetbrains.kotlinx:kotlinx-coroutines-bom:1.10.2
|         |         |    +--- org.jetbrains.kotlinx:kotlinx-coroutines-core-jvm:1.10.2 (c)
|         |         |    +--- org.jetbrains.kotlinx:kotlinx-coroutines-core:1.10.2 (c)
|         |         |    \--- org.jetbrains.kotlinx:kotlinx-coroutines-slf4j:1.10.2 (c)
|         |         \--- org.jetbrains.kotlin:kotlin-stdlib:2.1.0 -> 2.1.21 (*)

My original approach was to run this command and parse its output to retrieve our dependencies, but I didn’t like it. No real reasons to justify this except that it didn’t feel right. I then quickly realized that all the dependencies are in a version catalog. If you’re not familiar, the version catalog is a file in the TOML format that looks like this:

[versions]
ktor-version = "3.3.0"

[libraries]
ktor-server-core = { module = "io.ktor:ktor-server-core-jvm", version.ref = "ktor-version" }  
ktor-server-core-html-builder = { module = "io.ktor:ktor-server-html-builder", version.ref = "ktor-version" }

I don’t have anything that can parse TOML³ so instead I went for a naive approach where I read the file line by line, remove the comments and use a regular expression to retrieve which module we’re using and at which version. How confident am I in this approach? A solid 2 out of 10: what is easy to break is probably just as easy to fix, so that’s good enough right now.

Getting all the versions available for a module

The next step was to retrieve each modules’ available versions so we can compare with the version we’re using. After a quick search, I found out that there is a search API available for the maven central repository, so I started with that.

As the screenshot above shows using RapidAPI⁴, the API accepts a query (see 1) with a group (g), and an artifact name (a). Please don’t ask me what the core argument means, for I don’t remember. Using the filter’s feature (2) that accepts a jq syntax, you can see what we retrieve a list of versions (3). Good enough!

Unfortunately, I’ve found this API to be slow and unreliable, as most of my requests failed with timeouts. At this point it’s not clear to me if this search API is still actively maintained or not. I’ll try to find out because my alternative solution was a little, well, alternative.

Instead of using the search API, I’m grabbing the content of the page that lists all the versions available on a repo⁵ (in the screenshot above, those are the versions for the sql-gauss-db artifact in the com.easy-query group) and parse the HTML. That’s a little fragile, but it does the job for now. Plus, the chances that the format of the URLs on that page change in the foreseeable future are pretty low⁶.

Comparing versions

Finally, I have the versions we use and the versions that exist, so I originally started by sorting the list alphabetically and grabbed the top version as the latest. I knew this was a bad idea that I would have to fix eventually, but I didn’t expect I would have to address it immediately when my script suggested I upgrade a library from 1.12.x to 1.2.x.. As a result, I resigned into writing a proper semver parser (that doesn’t account for pre-releases versions yet) and addressed that issue.

The result is a simple script that shows the version used by a library and the latest version available upstream, in this simple format.

com.cjbooms:fabrikt                                : 24.0.0 -> 24.1.3
org.jetbrains.kotlinx:kotlinx-serialization-json   : 1.8.0 -> 1.9.0
org.jetbrains.kotlinx:kotlinx-datetime             : 0.6.1 -> 0.7.1
org.testcontainers:testcontainers                  : 1.19.8 -> 1.21.3
org.testcontainers:postgresql                      : 1.19.8 -> 1.21.3
...

Of course, there is already a bunch of things I want to improve, but this feels like a straight path toward re-implementing a bad version of dependabot, so I just merged my pull-request, selected a few dependencies to update first and moved on with my day, giving time for those improvement ideas to settle. I guess I’ll get back to it when all my dependencies are up to date?

We have family visiting, so we decided to rent a place on the Sunshine Coast. I have made myself a nice spot to work from and I’ve been taking the occasional break swimming, kayaking or standup paddleboarding¹, adjusting my time to enjoy the weather while getting things done. I also stopped by the local public library while the rest of the family is wandering around and it’s always a good experience. Once again, we really need to protect public libraries at all cost.

Since my favourite podcast has ended, I started listening to the first episodes again, like a ghost stuck in a loop unable to move on… until I discovered the Buffering the Vampire Slayer podcast. Despite everything we now know about Joss Whedon, the Buffy TV show was a huge part of me growing up. I watched it religiously when it was aired in France², then on repeat when we got the DVD Box Set. This podcast really hits the nostalgia part of my brain, the hosts are funny… I highly recommend it. I’ve also started to re-watch all the episodes, because why not.

I finally started reading The AI Con by Emily M. Bender and Alex Hanna and it’s a very good read so far. I’m also reading Legends and Latte, which is a lot lighter so overall it’s a healthy balance. Everything is terrible but we’ll always have giants orcs and coffee shops.

Finally, I put a fresh coat of paint on my blog. When I was trying to make some changes to include a link to my /now page, I realized that the stylesheet that came with the theme I setup a few years ago and tweaked over the years was out of control. So I did what everybody with a few hours to kill would do, I removed it altogether and re-did the integration myself. It’s not perfect, I still have some minor fixes to do here and there, but I’m glad I was able to do it. I’ll probably write a dedicated blog post about it soon³.

GitHub will always have a special place in my heart, it hurts to do the maths but I think I started using it in… 2008? I’ve been using (and supporting!) them ever since: I used the now defunct Job API as an example in my book, I contributed to libraries using their API, beta tested products like the iOS app or GitHub Actions, and I even tried to provide helpful feedback on Copilot, despite the fact that… well, you know.

That being said, this has been a long time coming for various reason, but these days it feels like your ethics is the only thing you can control, so I’m chosing to follow mine and leave. Last week, with the TV in the background, I gradually started removing stuff from there and it wasn’t as painful or annoying as I thought it would be.

First, I used gh (a tool I really like) to list all the forks I owned. The first half were a bunch libraries from my buddybuild days that I used to fork and try to onboard, while the other half were mostly Apple repositories that I forked once I was (finally!) allowed to do some open-source work. There was nothing really interesting nor surprising, but I’m pretty sure I’d done a pass on recent forks already.

The next thing I did was just going through all the repos I created. I went through a lot of various emotions: confusion for the empty repositories with a cryptic codename I could no longer explain, frustration for private side projects I worked on but couldn’t publicly release because of Apple’s open-source policy, and nostalgia for projects I’d simply abandoned. From there, I made decisions:

• the stuff I cared about but that provided little value, I simply downloaded an archive, deleted the repository, and added to my will¹
• the aforementioned stuff with cryptic names and often very little code, I simply removed
• the rest I pushed to my fresh Codeberg account, marked as archived on Codeberg and deleted on GitHub. It’s usually stuff that I mentioned in blog posts or side project I’m actually working on.

Codeberg has been fun to use so far, it feels like GitHub in the early days with a snappy UI and no constant mentions of an AI offering I simply don’t care about. I like the fact that it’s backed by a non profit based in Europe² and that it’s based on an open-source tool, Forgejo. I briefly considered self hosting my own instance for fun, but I know it would only be fun for a short while. I reserve the right to reconsider after I’m done packing food rations and building my nuclear bunker. In the meantime, I put my money where my mouth is: I downgraded my GitHub account (that I was paying for extra CI minutes) to free and gave that money to Codeberg.

I still have a few private projects to migrate, such as Comics Outmash and this blog (which uses Netlify). It’s only been delayed because I still need to figure out the CI situation. This is the kind of stuff I might consider self-hosting because access to Codeberg’s CI instance needs to be reviewed first, which I totally understand.

I don’t expect my decision to impact GitHub, but at this point it feels worse to put my ethics aside and embrace the enshittification, so I’m taking the leap. I guess we’ll see in 6 months if and how my life was transformed by this change?

I’m currently working (and have been for the past year or so) at Sound Games where we do… game stuff. I can’t really share much at this point, except that I’m in charge of making the backend go brrr. I never really saw myself joining a video games startup but after 7 years at Apple, I needed something different. I have already learned a lot of new stuff and the people are awesome.

Still recovering from a pretty shitty summer, but it’s hard to truly focus on that with family visiting. I’m expecting a lot of running in the upcoming weeks because I truly miss it and it’s been hard to commit to it when you’re trying to entertain, work and the weather is super hot. I do miss the people from my running club too, which is surprising since it involves waking up at 6:30 every Saturdays. We’re currently staying at a friend’s cottage on the coast, so at least I’m able to swim (albeit briefly because it’s f- cold) and go on short kayak trips.

I think this will be a good place to add what I’m playing, listening, watching and so on.