Lamdas don't capture function scoped variables

[joss-aztec]

Aim

Correctly compile constraints described in the body of a function, that are subsequently referenced in the body of a lamda.

Expected behavior

Circuit A (broken):

use dep::std;

fn do_lamda(x: Field) -> Field {
  let y = x + 1;
  let func = |z| {
    std::hash::pedersen([x, y, z])[0]
  };
  func(x)
}

fn main(x: pub Field) {
  constrain do_lamda(x) != 0;
}

The above code should describe a complete ACIR, with execution equivalent to this next program.

Circuit B (working):

use dep::std;

fn main(x: pub Field) {
  let y = x + 1;
  let func = |z| {
    std::hash::pedersen([x, y, z])[0]
  };
  constrain func(x) != 0;
}

Bug

The first code sample drops the constraints described by let y = x + 1; in the body of do_lamda. Here's the ACIR output:

current witness index : 7
public parameters indices : [1]
return value indices : []
BLACKBOX::PEDERSEN [(_1, num_bits: 254), (_2, num_bits: 254), (_1, num_bits: 254)] [ (_3,...,_4)]
DIR::INVERT (_3, out: _5)
EXPR [ (1, _3, _5) (-1, _6) 0 ]
EXPR [ (1, _3, _6) (-1, _3) 0 ]
EXPR [ (-1, _6) 1 ]

Note that the only references to Witness(2) appears in the pedersen input.

In contrast, here's the ACIR that's correctly generated from circuit B:

current witness index : 7
public parameters indices : [1]
return value indices : []
EXPR [ (1, _1) (-1, _2) 1 ]
BLACKBOX::PEDERSEN [(_1, num_bits: 254), (_2, num_bits: 254), (_1, num_bits: 254)] [ (_3,...,_4)]
DIR::INVERT (_3, out: _5)
EXPR [ (1, _3, _5) (-1, _6) 0 ]
EXPR [ (1, _3, _6) (-1, _3) 0 ]
EXPR [ (-1, _6) 1 ]

Note that the first expression decribes Witness(2).

To reproduce

1. nargo print-acir for Circuit A

Installation method

Compiled from source

Nargo version

nargo_cli 0.3.2 (git version hash: ad3e9fe, is dirty: true)

@noir-lang/noir_wasm version

No response

@noir-lang/barretenberg version

No response

@noir-lang/aztec_backend version

No response

Additional context

No response

Submission Checklist

• Once I hit submit, I will assign this issue to the Project Board with the appropriate tags.

[kevaundray]

This can be re-reviewed when closure conversion is implemented

[kevaundray]

@jfecher can this be closed?

[jfecher]

Looks to be working now