xyzzy_apps

Great observations on dynamic vs static typing. I often feel that there is a math envy between static vs dynamic types because dynamic types feel simpler as opposed to category theory. However the more I deal with dynamic types, its clear to me that the mathematical roots of dynamic types is in control theory. Its just a hunch that I have.

The human equivalent of LLM could be to build a wiki / semantic graph of all algorithms, tutorials, framework docs, stdlib docs and attach a make file at the end each node.

I've been stating it elsewhere, attention has made English into a context aware sensitive programming language by resolving ambiguities through anaphora and coreference resolution. Inform7 did limited anaphora resolution - it - but LLMs are able to resolve "move that file to this folder".

My bad, I completely forgot about Kotlin and Godot. Slint does seem promising same can't be said about Zig / D. When I mean supported it shouldn't feel hacky.

- C# by Unity and MAUI

- Flutter

- Qt

- Godot

- Kotlin

- JS

Personally I find my self sticking to Qt as the safest bet. Don't event get me started on VR support. Languages need to take Platform support seriously.

Every time I get excited to try a non C++ based language - Rust, Zig, D I am always reminded on how non-existent the mobile support is. Strictly from the standpoint of strong mobile support the only languages worth bothering with are C# and dart. Java and Swift are too tied to their platforms.

Absolutely!

DeepSeek feels like an excellent sparring partner. The others feel like mind readers.

I don't have to worry about any complex context engineering / multi agent stuff if I am using DeepSeek because the sequential stuff is fast enough. I don't have to bother about touching a million context length because its cheap enough.

Where it seems to lag behind is 3d, computer use, motion graphics, design but the initial demos seem like the vision model has improved substantially.

If painters have a canvas then programmers have a buffer.

The prompt and most reasoning is implicit inside the codebase. From every subsequent commit one can easily infer what the bugfix prompt was supposed to be or reasoning failures are.

All these repositories can easily be a part of the pre-training / post-training datasets. Perhaps the distillation the blog is speaking about is for text-generation not the code generation.

Isn't using all the repos uploaded by users to github a simpler way to distill ? Not only do you get Claude's intelligence but also the user's prompt guidance on top of it.