Max Rees

@[email protected]
38 Followers
42 Following
552 Posts
Posts are my own and do not reflect the opinions or policies of any organization or other life forms
Homepagehttps://www.cs.purdue.edu/homes/mcrees
Pronounsthey/them
InterestsC programming, systems integration, build automation
Max Rees22h ago
Edwin Brady1d ago

Today, thanks to a former PhD student, I learned that LLMs are terrible at generating Whitespace code: https://arxiv.org/pdf/2603.09678

Also today, I was invited (and I am accepting) to join a group on redesigning our undergraduate curriculum, which includes discussions on our choice of first programming language.

I am not saying that these things are related. I just happen to be mentioning them both at once because I saw them both today. You get a lot of characters in one toot after all.

What?

Show threadMax Rees23h ago
@xgranade @whitequark is part of the impedance mismatch here because mypy does a poor job of communicating that it is not necessarily type checking Python, but rather type checking for the purpose of mypyc? I think that at least explains the void-like return value example earlier.
Max ReesMar 13
Danielle ForéMar 13
LLM code is the lead pipe of our generation of software engineers. They’re going to be replacing it for decades and wondering how we were so foolish
Show threadMax ReesMar 12
@ariadne where can I connect to talk to this LLM. I want to see if it retained some vintage IRC memes
Max ReesMar 10
← Mar 9
Max ReesMar 10
Paco HopeSep 5, 2025
One of the ways that LLM-authored code improves productivity is by merely SAYING it does things. It's way faster than the whole time-consuming process of actually doing things. This is real code someone sent to me for review.
Max ReesMar 7
Cassandra is only carbon nowMar 7

"It writes boilerplate faster."

Is that all we aspire to now? Not abstraction or templating or better language design, just... output boilerplate faster.

It's depressing.

Show threadMax ReesMar 6
@lily I realize now that macports didn't have the good sense to store the git commit hash from which I built, nor was it embedded in the openconnect binary's version information... but judging by the date I must have cloned off master at commit 0dcdff87db65daf692dc323732831391d595d98d. The specific commit that supposedly fixed the problem is linked below, which I found through issue #659 that is linked in the commit message. I think I ended up cloning off master instead of just applying this one patch because it didn't apply cleanly to the last released version and I was too busy to tweak it. https://gitlab.com/openconnect/openconnect/-/commit/94e0b16c011b7b88708b8a8505fac6bfbe2e3cca
Use RFC9266 'tls-exporter' channel bindings for Cisco STRAP with TLSv1.3 (94e0b16c) · Commits · OpenConnect VPN projects / OpenConnect · GitLab

Fixes #659 Signed-off-by: David Woodhouse

GitLab
Show threadMax ReesMar 6
@lily a couple of months ago I realized I needed to build an unreleased version of openconnect due to some upgrade on the server side. If you're experiencing problems with the connection dying after every hour connected then this is why; it is a bug in opeconnect that only ~recently became a problem. I can share the specific commit needed if that is the case.
Max ReesMar 6
e. hashman 🇵🇸Mar 6
Someone said something about ingesting libc and this is what popped into my head