Restrict a Smart Contract function to be callable only from a specific Smart contract?

Question

I have 2 Smart Contracts: A & B. B has a function onlyA() which I want that only A can call. Below is my approach:

• Make B Ownable, then add the onlyOwner() modifier to onlyA().
• Add the interface of B in A so that A can call the function onlyA() from say callMe().
• Transfer ownership of B to A after deploying it.
• Now, user can call callMe().

My usecase:

I'm building a Protocol Smart Contract (A) that has an NFT Contract (B) linked to it. I don't want anyone to directly mint an NFT from B, but only by calling a function mintNFT() in the Protocol Smart Contract (A) which in turn calls the function onlyA() in B which internally mints the NFT.

Protocol.mintNFT() -> NFT.onlyA() -> ERC721._mint()

So I was thinking of using the above approach I listed. But I have to add the interface of my NFT Smart Contract to my Protocol Contract which I don't want to do. It just seems to me like an anti-pattern. It is considered bad practice to add interfaces like this?

I would highly appreciate any suggestions for an approach different from the one listed above. Also, any suggestions on the overall re-design of my Protocol Smart Contracts are also welcome.

Update:

Since both contracts are part of the same project, I was also wondering about simply importing the NFT contract in the Protocol Contract rather than defining the NFT interface. Which one is a better way, gas-wise & Solidity best practices wise?

Answer 1

Make B Ownable, then add the onlyOwner() modifier to onlyA().

You can just create a new modifier that would restrict the function so only A can call it, and pass A's address inside of B's constructor, no need to use ownable. Like

contract B {
address addressA;

modifier onlyACanCall() {
    require(msg.sender == addressA);
}
constructor(address _A) {
    addressA = _A;
}
function onlyA() external onlyACanCall {
    // ... 
}
}

But I have to add the interface of my NFT Smart Contract to my Protocol Contract which I don't want to do

This is extremely common in smart contracts, why would you want to avoid it?

Answer 2

Make B Ownable, then add the onlyOwner() modifier to onlyA().

You can just create a new modifier that would restrict the function so only A can call it, and pass A's address inside of B's constructor, no need to use ownable. Like