Well, it's now backed by S3 and I've made the code public for anyone interested: github.com/vinimdocarmo... 🦆

GitHub - vinimdocarmo/quackfs: FUSE-based DuckDB file system 🦆 FUSE-based DuckDB file system 🦆. Contribute to vinimdocarmo/quackfs development by creating an account on GitHub.

Went down on a storage rabbit role in the past weeks 😁 github.com/vinimdocarmo...

It is indeed! I’d only warn on the speed of the free machine types. Pretty slow depending on your use case. But I guess most workshops have small scopes anyways.

Even better, if you setup your project using devcontainers people will be able to easily setup it up locally using any supported editor ✨

What if you have an array and you want to transform it into a max-heap? Here you go:

Can do some basic @duckdb.org operations now 🤩 🦆

Here's the AI generated implementation by the way: gist.github.com/vinimdocarmo...

That's usually how I like to use AI for coding: to do things I don't want to do myself. Letting it do (i.e. to try to) everything is really unrewarding for me 🫠 I keep the resiliency high and still try and experiment with LLMs quite often though

I was solving a DP LeetCode problem in Python and I really wanted to visualize my decision tree structure in the terminal. I'm not a Python expert nor I wanted to waste my energy implementing a "pretty print" function. Copilot was disabled, but I turned it on and did what one must (ask the AI):

Had some progress today! I could write to a “virtual” file backed by Postgres. Since I’m using FUSE to simulate my file system ops, the file content is being persisted into Postgres instead of my OS file system 🎉

I mind less when it gives the precise, correct answer. But it's even more frustrating when it just spits nonsense.

My attention span is pretty short and every time Copilot takes much time to "think" I lose interest. I'm not finding working with AI much rewarding when it comes to coding. Maybe I should do some changes on my workflow? Not sure. I'll keep trying.

I didn't want to setup my Mac machine to make FUSE work, so I've decided to setup a dev container for VS Code and that's been giving me some headaches. Almost finished though. I only need to finish setting up Postgres 🤞

Differential Storage: A Key Building Block For A DuckDB-Based Data Warehouse - MotherDuck Blog Differential Storage: A Key Building Block For A DuckDB-Based Data Warehouse | Reading time: 11 min read

Today is Play Day for me and I've started building a Differential Storage using GoLang, something similar to what @motherduck.com does under the hood: motherduck.com/blog/differe...

new video! this datatype → [1, 2, 3] ← is a bit of a marvel. it can store items of any size, grow to any length, rarely causes performance issues, and almost every modern language has one. what is it, really? and how does it work?

I see. The reason I asked it was because we tried dedup in a block-level at WeTransfer, but it didn’t really yield good results since the data set is usually unrelated.

Great article! I see how dedup can help in a repo-level. But does it yield similar benefits between different, unrelated repos?

Even though the position and company was a great fit, I put my and my partner’s needs and long-term goals in first place. I’m relieved but at the same time I feel a bit of grief.

I passed the interviews, but unfortunately they couldn’t hire me in Amsterdam anymore due to policy changes. They still tried and I was aware London would be the second option. The offer turned out to be London-based and after much deliberation with my partner we decided to stay in Amsterdam.

No worries! Hope you're having fun 🛝

Care to share a bit more of what you're working on? 👐

I'm at this point learning more about that and thinking about implementing a naive custom allocator in C++

I’ve passed to final stage: a 30min call with an Executive. That’s one stage further compared to last time, so that’s a win already 🥳 Godspeed 🤞

Today I saw this expression for the first time while learning C++ `std::this_thread::sleep_for(200ms);`. I was suspicious. How is `200ms` defined? Is that dynamic? What?!

💡 TIL: you can define your own literals en.cppreference.com/w/cpp/langua...

Ghostty Ghostty is a fast, feature-rich, and cross-platform terminal emulator that uses platform-native UI and GPU acceleration.

Ghostty is out of private beta!

#ziglang

ghostty.org

Doodle drawn by hand explaining pipes in Linux

That’s how I best learn things 🎨

Well, we didn't spend time confirming that iCloud Drive was loosing files, so that's still an assumption. But getting that folder out of iCloud Drive sync stopped the issue 🤷‍♂️

He uses a MacBook and his projects folder was being synced with iCloud Drive. So every time any file was changed, iCloud Drive sync'ed them with remote. Problem is, `.git` and `node_modules` folders change quite often and iCloud Drive started to loose some files between changes. What a nightmare!

It was a simple React app, but often times the git history and remote config would just go crazy. npm start would also just stop working with some weird timeout errors reading files. It was like this for weeks, until I noticed something funny...

I'm mentoring a good friend of mine - which also happened to be my manager and CEO of a startup I worked for - and we experienced some real, weird bugs using node modules and git locally! I've never experienced those myself, so it was really hard to find out what was going on 👻

I was way more in ease in the second interview process. The first time I created so much expectations and it led to big frustration in the end. I'm glad I didn't give up though. Regardless of how it'll pan out, I've learned a lot!