Not actually a devlog #13
I won't promise to post more because it would be a lie. However, I'm happy to share some links and stuff I'm working on!
📖 Reads
- 🇬🇧 How Successful Companies Go Blind. Local optimums have been my worst enemies for... almost my entire career. They break trust and team communication, and as this article perfectly explains, they can even make a company fail.
- 🇬🇧 Lucas' Laws of Project Management. Articles stating laws are usually pure click/ragebait to generate views. Here it might be the case, I don't know. However, it's also pretty interesting to see flaws I've experienced (and continue to) in different companies.
- 🇬🇧 Why everything breaks at 150 people. Having documentation is somewhat hard, but keeping it up-to-date and efficient is hard as hell. Managing input sources to create cohesive documentation is a massive amount of work, and creating ways to navigate it easily is at least as hard. The article explains the challenge when the team grows and offers a solution to do it properly.
- 🇬🇧 How to Write an Effective Software Design Document. A nice approach to writing a design document, with an exhaustive list of subjects to cover. It offers a clear description for each potential section.
- 🇬🇧 Before you delegate, ask yourself these 6 questions. Delegating can be a tough decision. Sometimes you do it too late, sometimes too early. Wes is giving us a checklist to do it correctly and at the right time.
🚀 Projects
- ShowTracker, a KMP application to replace TVTime. I wanted to explore KMP again through AI-assisted coding. It's still in development and I'm the only user. Thus it's hard to tell if the implementation scales correctly in terms of UX/UI, and maybe the codebase is only relevant to me and the Agent 😅. However, it allows me to play with OpenSpec and tinker with systems/tools to improve my inputs. It also gives me (more) confidence in the safeguards actually put in place. For example, it SHALL follow DDD principles, pass TDD gates, let me review 3rd-party libs and so on. This part lets me learn about things like SKIE while bypassing some rituals I prefer to skip if possible (e.g. creating an application through heavy IDEs).
- tableronde, a pet project using UniFFI (yes there is a pattern in my projects 😬). This one has 3 goals: helping me move some data from Google Sheets to a Supabase DB instead, doing some Rust, and learning about another way to ship multiplatform code. This one is simple and should stay simple to let me focus on the shared core Rust. However, to add more constraints, I'm using it to try and see the appeal of Bazel. Some parts are specs using OpenSpec and delivered thanks to an Agent but the core thing and architecture layers are me-made. And let me tell you, I'm not convinced by this build system. I want to explore further before giving an honest (but fair) review of it. Same for UniFFI: I'm pretty convinced by the technology and love many aspects of it, but I want to go further.