Kei Lambda

autoImplicit and its consequences have been a disaster for the human race. It has greatly increased the dev experience of those of us who write "advanced" code, but it has destabilized refactoring, have made fixing issues unfulfilling, has subjected contributors to indignities, has led to widespread psychological suffering (in the maintainer side to physical suffering as well) and has inflicted severe damage on the ecosystem. The continued development of projects with it will worsen the situation.

autoImplicit is getting in a way quite a lot. It is actually doing more harm than good. Typo? Fuck you, now your code is silently broken at call-site. Forgot to update the old references in body? Fuck you, now your code is silently broken at call-site. It might have been useful for the person who was writing the initial code. But for refactoring it is a nightmare. I don't know who in their right mind did not turn it off in the standard library code. Mathlib folk turned it off ages ago. 🤷‍♂️

Some more progress on this: diff

Maturity and sympathy cure envy.

Business is booming on twitter too. That's one of the reasons I stopped posting here frequently.

I had a productive argument for the first time on twitter too! The thread: https://nitter.net/khoiiiind/status/1895317796969685386

EA made a power move: https://github.com/electronicarts/ Command & Conquer series is now open source. There are so many possibilities to improve upon... we can even kiss goodbye to lags.

https://www.omocat-shop.com/collections/omori/products/mari-plush https://www.omocat-shop.com/collections/omori/products/space-in-between-enamel-mug https://www.omocat-shop.com/collections/omori/products/something-sling-bag I am a financially responsible adult. I am a financially responsible adult. I am a financially responsible adult. I am a financially responsible adult. I am a financially responsible adult. I am a financially responsible adult.

I deviated from the book many times and by a large margin - by choosing Haskell, by changing the conclusion block (added platform specific exit syscalls which were not present in the book), by writing my own tests and so on. Now thinking about deviating from the target language and platform. Maybe it is going to be good to adopt Intel syntax instead of AT&T and target ARM? Don't know, not sure, can't decide.

Finished the second chapter of Essentials of Compilation and assembled the whole pipeline: Pipeline.hs Compared to my other, abandoned, Lean 4 implementation this one is much more correct and more performant - both the compiler and the output of the compler.

By the way, today marks both Logic Day and Yukio Mishima's centenary.

Recruited another dev at our company into the Alloy 6 gang.

"If You’re Thinking Without Writing, You Only Think You’re Thinking" Why does every quote of Leslie Lamport go so hard?

Beauty will save the world? Why, yes, type theory is beauty and it will save the world.

Working my way through the book "Essentials of Compilation: An Incremental Approach in Racket" with Haskell: https://github.com/keilambda/eocia-haskell/ The interesting thing is, I found bugs on my previous iteration with Lean 4. Namely lexical scoping issues. So many of them. I look back at that time and wonder how could I write such bug ridden code. Learning Lambda Calculus (oh those substitutions, nightmarish substitutions and beta reductions) was of great help after all.

Full retard mode: second iteration. 17 hours 56 minutes straight in front of the computer.

I'll never be enough. In the end, I'm just a stupid back-end javascripter and nothing more.