Alasdair Allan
I built a programming language you're not supposed to read, https://veralang.dev. Vera is designed for machines to write. No variable names — typed De Bruijn indices instead. No style choices. Every function contract verified by Z3 SMT solver. Compiler errors are fix instructions for the model, not diagnostics for you. Pure by default, all effects explicitly declared and tracked. Point a model at it. #AI #LLM #OpenSource #Agent
https://negroniventurestudios.com/2026/02/28/a-language-designed-for-machines-to-write/
Vera — A language designed for machines to write

A programming language where verification is a first-class citizen. Mandatory contracts, algebraic effects, typed slot references, compiles to WebAssembly.

Mar 1 at 2:03pmWeb
Show threadEsteban Küber Mar 1

@aallan
> Traditional compilers produce diagnostics for humans. Vera instead produces instructions for the model that wrote the code. What went wrong, why, how to fix it with a concrete code example, and a reference back to the specification. This way, the compiler’s output is designed to be fed directly back to the model as corrective context.

How is that different to what all compilers should be doing for humans?

Show threadAdrian CockcroftMar 1
@ekuber @aallan It sounds more verbose, but your work on Rust error messages is the best I’ve seen so far in other languages.
Show threadEsteban Küber Mar 1
@adrianco @aallan thank you! From what I saw it reminds me of what I've seen from Elm. I think that verbosity can be dealt with by having toggleable levels (so you can reduce the amount of repetition and guidance that has already been given). For humans, interactivity is even better (being able to expand output and ask for more information without having to recompile). The only concern I would have here is coverage of errors on the tail end: the more complex or uncommon the conditions, the more of an explanation you need to give, but they are also the cases less likely to have had time spent to make the output more verbose.
Show threaddavecb 🇨🇦Mar 1
@aallan I rather like that idea. I wonder what meta-language will grow up to instruct the LLMs. I find english-language prompting even less precise than common programming languages.
Show threadJan WedelMar 1
@freakshow @Roddi
FYI
Show thread🧊 freezr 🥶Mar 1

@aallan

Very impressed by the philosophical and investigative approach!

Show threadAdrian CockcroftMar 1
@aallan I was wondering when something like this would appear. Nicely done, it would be interesting to try out my github.com/brazil-bench test to see how it compares to Python.