'EXTCODEHASH' opcode

[Arachnid]

This EIP proposes an 'EXTCODEHASH' opcode, which returns the keccak256 hash of a contract's code.

[jochem-brouwer]

@Arachnid

If I call extcodehash on an empty account it pushes bytes32(0) on the stack - which is according to the specification.

However, if I now create an account and then selfdestruct it (in the same transaction) I would suppose that after creation of the account it returns the hash of the code (which it does). But after I selfdestruct it, I would suppose that the hash is now either bytes32(0) since the account is destroyed or it is another hash than the hash which got pushed before I selfdestructed it (since the code is now empty). However, the same hash gets pushed on the stack. What is the reasonable for this? Does the code in an account get emptied after a transaction?

Please see this ropsten account:
https://ropsten.etherscan.io/address/0x1b3ebbd2a941e38db6f274a905581a4664dd9708

(array added for clarity)

I expect the 0-index of the array to be bytes32(0), the 1-index to be the actual hash and the 2-index to be bytes32(0) again.

[fulldecent]

There is a bug in 1052 and it should not be included in Constantinople in its current form.

I submitted it for the Ethereum Bug Bounty. And I didn't read the details of whether I am allowed to fully disclose details while it is being considered for bug bounty, so I won't say anything else here yet.