When we want to verify our proof with Integrity there are no issues if the proof was generated with the following layouts:
  • recursive
  • recusive_with_poseidon
  • dex
  • small
  • starknet
  • starknet_with_keccak.
But if we wanted to verify proof that was generated with layouts dynamic or all_cairo we are going to run into issues. Those layouts are not supported by Integrity, because of complexity of their verification.

​
Double Verification for Integrity Compatibility

As a workaround we came up with sort of “double verification”
  1. Firstly, generate proof for initial program using desired layout like dynamic. This will allow proof to capture full execution of the program.
  2. Then generate trace, proof and verify it, but instead of providing your program as programFile, use StarkWare’s verifier written in Cairo0 and provide proof from step 1 as an input.
You need to modify proof.json you got from proof verification to have object with proof key instead of direct object proof, so for example if you have proof.json like this: 
{
    "annotations": [...],
    "proofHex": "0x...",
    ...
}
you need to modify it to look like this: 
{
    "proof": {
        "annotations": [...],
        "proofHex": "0x...",
        ...
    }
}
As a program you need to use Cairo0 verifier or simply use 0x193641eb151b0f41674641089952e60bc3aded26e3cf42793655c562b8c3aa0 program hash. In the flow described above we are not directly verifying your program. Instead we proof that StarkWare’s verifier when executed on your proof passed verification. This indirect verification through double proving ensures compatibility with Integrity’s layout limitations. Note: This is only necessary when verifying with Integrity. If its not required in your case you can skip this and proceed with standard verification.