He spent his final few months knowing where he stood and how much he was loved. He served humanity and was held up until his final moment. May we all live this way and be so blessed. 🙏

rest in peace, Scott Adams

"In my experience, X works in situation Y, so we should do that." Like such things are knowable... I often falsely assume I know why the things I do work.

You shouldn't cargo cult yourself. At work we have some ideas of what a good process is. But considering introducing 10% "don't follow the process" time. For example, we do pair programming, not PRs. But maybe we should do _some_ PRs still, just for the sake of

Having a separate back-end and front-end yields microservice cancer :( We have to remember to push code properly and push code for the back-end first and wait for it to deploy etc, since they are built and deployed separately.

"Separation of concerns" wrongly assumes we have a good grasp of what our concerns are in code.

I really like @ladlejs. Weird how it gets almost no attention. It has every feature I need from Storybook, but it's much much simpler and is just a single dependency.

We just decided to re-monolith our codebase. Our background patch processing system is in a separate module in a monorepo. During a discussion about how we should port backend code into the common module that both projects share, we realized: why isn't it just one module? So,

GM

Now, all users secretly run this in prod (an admins see the debug info up in the top bar as well) Thanks, AI!

New AI experiment: have it set up a basic SSE architecture between our self hosted Next.js frontend and Kotlin/Ktor backend. An initial POC seems like a good fit for AI - I'll iterate on it myself and enhance it, but I haven't set up SSE this way (or really any way) before so I

To imagine that this was the web in the early 2000s

Getting into a good groove with the June coding agent in IntelliJ. Every time I tried to make it tackle complex business logic, it has failed spectacularly. But for doing relatively menial tasks, it really excels. Even for large tasks. As long as they are sufficiently menial.

Getting 50 top up credits just like that is probably nothing, but feels bull to a newcomer like me!

Me, every standup these days. Golive is March 2nd, wish us luck!

Just added @ladlejs to our frontend. I've never actually been on a project where we've managed to successfully maintain a list of stand-alone components. BUT THIS TIME WILL BE DIFFERENT. I hope. This also helps us enforce GUI code with absolutely zero business logic. Even in

Last night I worked until after midnight and slept in the office. Definitely not stable and long-term. Planning to sleep normally tonight. But if I can do it at LAN parties, why not do it at work every now and then? :D

At this point, I want us to take on as much technical debt as possible. At least in terms of "debt is good, because we can pay it later without much consequences" Difficult to actually manage this, though. What technical debt is difficult/impossible to fix later? What can we

DAE remember when we used to make fake packaged cardboard boxes for download only software? I made a bunch of these for Joomla templates (!), unfortunately they're lost in time.

AI agents has the same async problem as PR's: You review a bunch of code when it's done. We don't do PR's. Instead, we pair program and do the initial design synchronously. This lets us hash out design issues up front, instead of wasting hours or days on the wrong approach.