Álvaro Rodríguez

Ah true, forgot about the serialization change. I don't think it's released yet!

Oh, it went in in 0.28. Will try updating, let's see https://github.com/AztecProtocol/aztec-packages/pull/5192

Closing to try to fix sync

Shouldn't possibly mutated also look for arrays sent as parameter to call functions? The nested call might return it aliased even though it's a reference to the same array? :thinking:...

Ahh true, i missed the part where it did it by type and not by id! Will do a full review now :)

Oh I completely made up the size of the loop, it was just to show a functionally equal example so I used a loop that runs only one time. We...

I know, it's fine for it not to be unrolled (even in the case the bounds are known at compile time, unrolled loops can lead to bigger bytecode), I meant...

I'm looking at the SSA and it seems that mem2reg is not identifying that the loads are known: With loop: ``` After Inlining: brillig fn main f0 { b0(): v7...

I guess the ideal SSA mem2reg could output in this case given that It's not its job to remove loops with empty bodies, is something like this: ``` After Mem2Reg:...

This might be related, poseidon is a bit unusable right now in brillig: This code: ```rust unconstrained fn main() -> pub Field { dep::std::hash::poseidon2::Poseidon2::hash([0; 2], 2) } ``` Generates this...