Prevent contract name collision with @openzeppelin/contracts

[montyly]

Hi,

🧐 Motivation
@openzeppelin/contracts-ethereum-package and @openzeppelin/contracts share many contracts names. The duplication prevents the correct compilation artifacts to be generated if a codebase uses both packages.

📝 Details
Most of the compilation frameworks do not handle well codebase with contract's name duplicate. For example, when generating the json artifacts, Truffle will not be able to generate correctly the files trufflesuite/truffle#1087.

As a result, it can prevent the integration with third-party tools. For example, we had to develop a specific detector in Slither to know what contracts names are reused. But if a codebase uses both @openzeppelin/contracts-ethereum-package and @openzeppelin/contracts, it is difficult for a user to avoid the duplicate names.

Example
A codebase that would have both an upgradeable ERC20 token a non-upgradeable one will be affected:

// contracts/UpgradeableERC20.sol
import '@openzeppelin/contracts-ethereum-package/contracts/token/ERC20/ERC20.sol';

contract UpgradeableERC20 is ERC20 {
}

// contracts/NonUpgradeableERC20.sol
import "@openzeppelin/contracts/token/ERC20/ERC20.sol";

contract NonUpgradeableERC20 is ERC20 {
}

Running truffle compile on a codebase with these two files will result in one build/contracts/ERC20.json file that will either represent @openzeppelin/contracts-ethereum-package/contracts/token/ERC20/ERC20.sol or @openzeppelin/contracts/token/ERC20/ERC20.sol.

A better practice would be to have separate contract names according to the package.

Let me know if you have any questions,
Thanks!

[frangio]

Thanks for this suggestion! It's a good idea and we're going to consider this for the upcoming 3.0 release.

[frangio]

This has been partially fixed in 3.0 but interfaces still have names without suffix. We'll have to correct this in a later release.

[frangio]

Fixed in @openzeppelin/contracts-upgradeable@3.2.0.

See First release of OpenZeppelin Contracts Upgradeable.