I’m a programmer in Edinburgh. I write here mostly about programming languages and their features; types, laziness, syntax design, and how little evidence we actually have for the things we believe about them. I think many of my posts do not take a strong position or if they do I end up arguing against it.
A C++ browser engine is outpacing a Rust one, this goes against my priors that stronger guarantees are generally good for programming so I listed other reasons why this might happen, but I wasn’t convinced by any of them.
Function length is the wrong proxy for the thing we actually want.
Probably not absolutely nothing, but I think weakly typed languages are mostly useless, so I try to steelman the case for one
Why you can write a generic admin panel in Python and not in Elm — explained down at the level of what’s actually in memory.
A good example of what I think laziness is good for, allowing natural structuring code
The follow-up where I explain why laziness is not a silver bullet that always works.
Extensible union types are often proposed. I think the important difference is between structural and nominal types, and that this should be uniform across records and custom types.
Our F1 prediction scoring let a randomly-generated guess beat a nearly-correct one. Here’s the system I built to fix it.
Before proposing a feature, show standing. An under-rated analogy from Dillon Kearns.
I go through a podcast episode I was asked to listen to, and find that there isn’t much substance to the arguments.