branislav

I’m curious, how do you handle authorization of the user without leaking tokens/api keys to model context without an MCP server (which supports OAuth, in principle)?

I echo the other responses to your question. Just recently I had two instances where an old blog post of mine contained a solution to a problem a coworker encountered. I publish these notes mostly so that I don't forget about them, but it's an added degree of satisfaction when it proves helpful to others. I wrote a bit about this in one of my recent posts [1]. Don't be afraid of the rabbit holes!

[1] https://branislavjenco.github.io/rabbit-holes/

Controllers are a complex topic, but as the linked talk describes, it all comes down to some basic control theory concepts. I wrote about them in my Desired state systems post https://branislavjenco.github.io/desired-state-systems/ if somebody wants a high-level overview of how to think about them.

Basically, declarative state implies value semantics which makes it easier to reason about. Underlying complexity is high though, and you need to judge how necessary it is.

I’ve found that getting traffic through organic search isn’t that difficult, if you have a post which is quite specific. For example, some years ago I wrote down how to upload assets to an already existing GitHub Release [1] as a small note to myself, so that I remember it next time. That is one of my best performing posts, majority of traffic via search engines, and I didn’t advertise it anywhere.

It by no means gets thousands of views per day, more like single digits, but people keep finding it, which gives me hope it’s been useful for others as well.

[1] https://blog.br4.no/github-actions-release/

One way to look at React is as an open loop desired state system, where you have a declarative way (js/jsx) of describing what you want to render and the reconciler figures out what operations to do (in the DOM), ideally in an effective way. But you could swap out the render target to other things. I wrote about it some time ago https://branislavjenco.github.io/desired-state-systems/

I also highly recommend this [1] episode of the Omega Tau podcast which touches on black holes, holography and quantum gravity.

[1] http://omegataupodcast.net/191-string-theory/

It's also how something like React.js works. Wrapping stateful DOM manipulation in a functional shell, made performant thanks to the Virtual DOM diffing algorithm.

Incidentally, I am on Xubuntu, and I would love to have the workspace switching transition that macOS has. Switching workspaces on xfce is lightning-fast but I would love to have an indication on whether I moved left or right. (other than the small indicator in a Panel) Ubuntu solves this with a HUD (or at least did with Unity).

I agree with the article in that designers can be too focused on the shiny and non-functional aspects of interfaces.

Personally I would treat design as being purely about function and anything superficial is about aesthetics.

But I think the author is throwing the baby out with the bathwater by wanting to restrict all animations. Fast, subtle animations, especially in the form of transitions between states/screens are extremely important for keeping the user informed about where they are in the context of the system.