ここしばらくは空いてる時間でキツネという分散型SNSを開発している。分散型SNSは一台のサーバーで完結せず、ActivityPubやOStatusといったプロトコルでいろんなサービス（Mastodon, Pleromaなど）とのサーバー間の通信も重要。
本番に上がって使ってもらう時はもちろん、現状の開発中でもログは不可欠。バグがもしあった時に、いちいちデバグ文をあちこちに入れて何とか原因を突き詰める作業をできるだけ減らしたい。俺に考えつくのは、例えばリクエストや処理が失敗したとしても、初段の調査に最低限必要な情報は既にログにあるようにしたい。
The other day I got an email from Axosoft that the payment for my GitKraken license (yes I use GitKraken) was rejected by my credit card company. It was because the card I used there expired, but nonetheless I had to take action.
That involved going on their website and changing my payment method to a card that still worked. It involved first an overwhelmingly complex table like this.
When trying to pass data between a Docker container and the host, using ADD in the Dockerfile might be sufficient at first. However, it’s one way, get burned in the image and so very inflexible.
The usual solution is to mount folders using docker‘s -v option. It’s simple, easy to use and pretty reliable. Just add -v "$(pwd):/root" and the current folder will be mounted to the /root folder in the container.
Using volumes is nice because they’re (can be) two way and (can) sync in real-time. Now you don’t need to rebuild your image every time you fix a typo. -v has pretty deep configuration options too, in case you want to go down the rabbit hole.
A while back I was trying to implement HTTP signatures to use with ActivityPub interactions with Mastodon. In Clojure. There is a go-to library for Clojure when it comes to crypto stuff, but I couldn’t get it to do the specific thing I needed: SHA-256 / RSA signatures. I looked at other options too, but as I’m not familiar with NaCl, that was just a confusing mess of wrappers around Java wrapped around C.
In the end I went with using Java interop to call Bouncy Castle stuff directly. I hate Java and interop in Clojure just feels wrong, but at least I could get it to work. Not to mention if something, Bouncy Castle is maintained. It wasn’t exactly a joyride, but it works. Check out the source if you’re interested (or want to use it). I didn’t make it stand-alone or put it up on Clojars (yet).
Recently I’ve been working on a Clojure implementation for ActivityPub. In the process I wanted to use specs, but I ran into a pretty significant problem. Namely the very first line in basically every single ActivityPub JSON object: { "@context": "https://www.w3.org/ns/activitystreams" }.
Do you see the problem? Well. This JSON will arrive at the server, where it’ll be handled by Cheshire or something along those lines. Point is, keys in JSON maps will end up turned into keywords. Clicked the link? The guide isn’t exactly specific about what can and can’t go into a keyword.
This weekend I decided to upgrade my old Android #nowplaying app. For the new version I decided to use Kotlin and also make it usable from my Android smartwatch. Honestly, it went much smoother than expected, except for one big caveat.
Out of some weird impulse, I decided to target the Oreo SDK (26). Both my phone and my watch are Oreo based, so that wouldn’t be a problem. Yet I kept running into the weird problem that my BroadcastReceivers wouldn’t receive any intents whatsoever. Interestingly enough, the WearableListenerService I used to to listen to Messages from the watch (I want to post when I tap a button on the watch) worked almost instantly.
I originally planned to make a Rust program that would fix the broken encodings in my music files. Luckily, I could solve that without Rust. Luckily, because from what I’ve experienced since then, it would’ve been a horrible pain in the ass to do.
I still wanted to give Rust a try. Contributing Servo is pretty much out of question – it’s just too high level to jump into as a complete beginner. (And honestly, I don’t think I’ll get past “complete beginner” with Rust anytime soon if ever.) It wasn’t hard to find another issue.
Become Immortal is at this point the only 1kyu kata I tried (and solved) on CodeWars. It took me 3-4 days to solve it. It was definitely pretty damn difficult to figure out, but in the end I managed it myself.
Looking at the example tests it was immediately obvious that trying to generate the actual rectangles and xoring everything was out of question. For the heck of it I tried once, but even for the last “basic test” I’d run out of memory (on my desktop) very quick. Sure it works fine if the rectangle is 30-40 rows and columns, but not for scales in the millions. I used it to dump rectangles on the console and double check my solution.
Another CodeWars kata, by the same person who did Twice linear. It was annoying in a different way. Once again, a naive solution was out of question (or so I thought), so I kept trying to figure out a way to shortcut the whole problem with maths.
Generating the integer partitions (and their products) isn’t particularly difficult. There are plenty of resources out there about how to do it – though for some reason mostly in Python.
There’s a CodeWars kata called Twice linear. It’s one of those problems where a naive solution is just not feasible. Not to mention the first tests aren’t indicative of what’s expected so by the time I hit the wall of very big numbers, I already had a solidified (and naive) mental model. Getting rid of that isn’t an easy task.
Generating the list of numbers isn’t difficult. Keeping it sorted and not adding duplicates makes it a bit more tricky. I tried using Ruby’s SortedSet at first, then Enumerators and plain loops, but everything crapped out at bigger numbers. All tests are supposed to run within 12 seconds on CodeWars, and my code at that point was taking 20+ minutes to run for 60000.
七大陸最高峰チャレンジCopyright © 2025, valeblog. Use for AI training forbidden. Proudly powered by WordPress. Blackoot design by Iceable Themes.