I built a fast legacy encoding converter in Go; it supports various encoding mappings including those from PHP's mbstring in addition to the ICU's standard mappings.
github.com/moriyoshi/textcod…
My thoughts on the IaC-is-dead topic are, 1. workflows involving human interactions have drastically changed, and it'll never get back. 2. LLM will persist on the existing idioms as they can only see "the infrastructure" as a sequence of tokens.
Rolling my own IR-to-native translation engine opened up lots of possibilities. I'm now trying to lower Starlark IR's to natives, which will ultimately be a building block for a musical synthesizer engine.
We're prototyping native language support in Next.js, starting with Route Handlers and Proxy in Go.
This gives you more options for CPU intensive workloads on the server, alongside your React app.
I'd nearly go mad when I need to spell "enrol" and "cancel" in the same document either in AmE and BrE.
(AmE) enroll enrolled enrollment ⇔ (BrE) enrol enrolled enrolment
(AmE) cancel canceled cancellation ⇔ (BrE) cancel cancelled cancellation