latenightcoding

I'm sorry but your partner sounds like a pain in the. History is History, Go find yourself some ancient goods.

They have all felt the same to me the last decade, I'm not sure what I am using at the moment.

The Mojo PR team on their way to hire OP to publish these sort of articles at Modular.

Hey hey they are doing way more than that. They are also trying to change the national anthem… again!

Never seen anybody refer to chatgpt code as "codegen".

I genuinely wonder if tech companies who cave in keeping WFH for good are just going to start hiring cheaper talent oversees.

I’ve seen way too many founders succeed by pivoting to something completely unrelated once they have funding and a big team to hate on people who do this.

Structure and Interpretation of Classical Mechanics published in 2001 uses Scheme (There is a second edition released more recently)

Fast ai had a computational linear algebra class iirc

Good marketing: it’s not even a simple language but people marketed it as executable pseudo-code and it worked.

Also Perl dropped the ball with Perl 6.

With sublime I can have 5-10 projects open without even noticing (LSP plugin and everything). vscode can barely handle one.

Good to know, it never sat right with me that Guile depended on BDW-GC for so long. That's what you use when you are implementing a toy language and don't have time to write your own GC.

or category theory phase

Classic paper but what the hell is this title .

Read your article (and cloudfare's) and as someone who uses musttail heavily I don't understand the hype, as you mentioned in your blog: you can get tail call optimizations with (-O2), musttail just gives you that guarantee which is nice, but the article makes it sound as if it unlocks something that was not possible before and interpreters will greatly benefit from it, but it's more reasonable to ask your user to compile with optimizations on than it is to ask them to use a compiler that supports musttail (gcc doesn't). Moreover, musstail has a lot of limitations it would be hard to use in more complex interpreter loops

being into AI at the right time and the MIT affiliation helped.

It's actually impressive Lex is this bad at something he has been doing for so many years.

ML people nowadays barely know basic stats.

A literal hacker news. RIP.

So what happens next? does Stanford and his past employers sue for what he has already been paid? Does he lose his prestigious degrees? Because if this ends with his resignation, the fraud was totally worth it.

I am giving them an out.

title could be: "nerd tunnel vision" lead me to work in a Ponzi scheme.

Many many years ago Ethereum /smart contract hackathons way were more innocent, at least most participants acknowledged they were working on something useless but fun.

A Ponzi scheme still a Ponzi scheme even if it's using Rust.

I just want auto-renew to work properly sometimes it works sometimes it doesn't, lost a couple of high value domain names even tho they've had my CC for years.

Sounds like something cooked up by a fradulent crypto bro who doesn't know anything about law or finance, but it's reality lol.

And you can't hide who you are following (even tho you can on fb), which is a big deal.

>> Ignoring the fact that you incorrectly refer to LLVM IR as LLVM

I also "incorrectly" wrote "target C" instead of "target C code". It's obvious what I meant.

>> Comparing the 2 is nonsensical to say the least.

You could have easily googled "C vs LLVM as a backend" to find a lot of interesting discussions comparing both before being so confidently incorrect

It does make sense, your language can target C or LLVM. each option comes with pros and cons e.g: C is more familiar, LLVM has better support for JIT compilation.

Something that I find a little concerning about Racket is that AFAIK there are no alternative implementations. (Clojure and Common Lisp have a couple). Why is that? is it too complex?

"unveiled at North America’s top tech conference" ... Collision. Give me a break, Collision is Grifter Olympics.

“Rule 4: When you want a raw pointer as a field, use an index or an ID instead.”

literally just woke up but: wouldn’t it be simpler to use a pointer to a pointer, or am I missing something