solving ZK Hack V puzzle 3

My first impression of Noir from the short puzzle alone is that it’s clean. I haven’t had much experience with writing Circom but reading it is a pain.

Error messages in Noir are also far more useful compared to in Circom – if being Rust-based has had any positive effect, it would be the helpfulness of the compiler and the tooling. Speaking of tooling, nargo is great!

Now, the not so good. These are probably just nitpicks at this point because I haven’t had enough experience with the language to give a fair criticism, so take these comments with a grain of salt.

I think the way inputs get fed into the Noir program doesn’t seem right. Firstly, it’s not super clear that inputs are implicitly fed from Prover.toml, and secondly (this is more of a nitpick), TOML is a terrible format for such a purpose.

I think JSON would be more suitable here for a clearer hierarchy of inputs with nesting - no more awkward “table of tables”. Especially for the purpose of just passing in input to the program, TOML is overkill and is suited for more verbose purposes eg. configuration like in Cargo.toml. But I guess at least YAML wasn’t chosen here.

Another ‘feature’ I’m cautious about is the fact that it is a Rust-based DSL. Rust looks like Rust because of its design choices. I don’t think languages should emulate Rust just for the sake of ‘familiar syntax’, to capture Rust developers in the space. I don’t think this is the sole reason for doing a Rust-based DSL, but it’s definitely one of the reasons that people often cite as a positive. I’d argue that Rust syntax is overly verbose and ugly. It only looks that way because of the amount of work that is frontloaded to the compiler. Especially for circuit development, (I believe) there is no need to be ‘Rust-like’ at all, unless you’re reinventing rustc, except this time for circuit compilation.

That said I don’t have too much else to say atm. I will have to write more Noir before I figure if I like it enough or not.