Record field ordering issue in shared term type checking

[atomb]

When running the ECDSA proof with extra type checking enabled (the -t flag), I'm noticing some failures due to differences in field ordering in records. For example, the following is one of the messages (edited slightly for readability):

Function argument type
  let { x0 = #{};
      x1 = Prelude.Vec 384 Prelude.Bool;
      x2 = #(x1,x1);
      x3 = (x :: x1) -> x1;
    }
 in #(#{"add" = (__p1 :: x2) -> x1 |
      #{"div" = (__p21 :: x2) -> x1 |
      #{"field_unit" = x1 |
      #{"field_zero" = x1 |
      #{"half" = x3 |
      #{"is_equal" = (__p22 :: x2) -> Prelude.Bool |
      #{"is_val" = (x :: x1) -> Prelude.Bool |
      #{"mul" = (__p3 :: x2) -> x1 |
      #{"neg" = x3 |
      #{"norm" = x3 |
      #{"sq" = x3 |
      #{"sub" = (__p2 :: x2) -> x1 |
      x0}}}}}}}}}}}}
     ,#{"x" = x1 |
      #{"y" = x1 |
      #{"z" = x1 |
      x0}}})
doesn't match expected type
  let { x0 = #{};
      x1 = Prelude.Vec 384 Prelude.Bool;
      x2 = #(x1,x1);
      x3 = x1 -> x1;
      x4 = x2 -> x1;
    }
 in #(#{"is_val" = x1 -> Prelude.Bool |
      #{"norm" = x3 |
      #{"add" = x4 |
      #{"sub" = x4 |
      #{"neg" = x3 |
      #{"mul" = x4 |
      #{"sq" = x3 |
      #{"half" = x3 |
      #{"div" = x4 |
      #{"field_zero" = x1 |
      #{"field_unit" = x1 |
      #{"is_equal" = x2 -> Prelude.Bool |
      x0}}}}}}}}}}}}
     ,#{"x" = x1 |
      #{"y" = x1 |
      #{"z" = x1 |
      x0}}})

Modulo naming of type parameters and reordering of fields, these seem the same to me.