Anders Hejlsberg is the creator of TypeScript and C#, and I asked him about how the TypeScript compiler got 10x faster through a rewrite in Go and his though...
00:00 Intro
00:48 Why write a compiler in JavaScript
07:29 Why rewrite the compiler in Go
14:49 LLMs for large migrations
20:12 Why Javascript is so popular
26:32 Why ever use Javascript on the backend
32:59 What it takes to build a programming language
37:06 Will there be fewer languages in 10 years
42:57 Hands on engineering vs delegation
49:14 Why fast tooling matters more now
51:16 AI software engineering predictions
58:52 The most technically challenging work
01:02:04 Top book recommendation
01:03:50 Advice for his younger self
01:05:00 Outro
Just summarizing the title teasers:
10x faster by switching from JavaScript to Golang. Rust was not feasible because it does not have a GC/does not support circular dependencies, and they have many of those, and did not want a rewrite but a minimal effort switch.
AI may produce a lot more code, but what code does “90% of code” refer to. AI is not capable of novel or complex stuff, you still need humans in the loop. Juniors remain significant to train them into seniors.
This seems to be both untrue and mostly irrelevant.
The recent FLT proof is clearly complex and novel.
But also a huge amount of programming is only barely complex and not really novel at all.
The biggest issues I’ve seen with AI so far are:
The slop comments are just painful to read. I think this might mostly be Claude though, I recently switched to Codex and it’s way less annoying.
It is so benchmaxxed to “succeed at all costs” it will often do stupid things just to make the test pass or whatever. To be fair a lot of humans are like that - “I dunno but it seems to work” attitude.
Just summarizing the title teasers:
10x faster by switching from JavaScript to Golang. Rust was not feasible because it does not have a GC/does not support circular dependencies, and they have many of those, and did not want a rewrite but a minimal effort switch.
AI may produce a lot more code, but what code does “90% of code” refer to. AI is not capable of novel or complex stuff, you still need humans in the loop. Juniors remain significant to train them into seniors.
Thanks for the summary! Initially I thought the headline was about writing Typescript programs 10x faster.
Thanks!
This seems to be both untrue and mostly irrelevant.
The recent FLT proof is clearly complex and novel.
But also a huge amount of programming is only barely complex and not really novel at all.
The biggest issues I’ve seen with AI so far are: