Rename/alias sha3 to minimize confusion with SHA-3 standard

[ethers]

We should rename or have an alias for Ethereum sha3 to something else, to minimize confusion with the SHA-3 standard. Even though Yellow Paper and some docs mention Keccak-256, having sha3 in Solidity code is highly misleading. Some suggestions are ksha3, keccak. Any others?

Aliasing say ksha3 to sha3 (preserves backward compatibility) and encouraging the use of the new term, and then deprecating sha3 in a future version of Solidity, seems like a practical option to move forward.

[axic]

I think there are multiple separate instances of Keccak/SHA3 usage and a single approach to all might not suffice. It is used at least:

• as an opcode in EVM
• an instruction in Solidity/etc
• as part of the address, tx, block, etc. hashing
• in the ABI
• in ethash
• in the Merkle Patricia trie for hashing the keys

It probably is unrealistic changing Keccak to SHA3 in all the above. It is fairly simple introducing a new opcode in EVM and it is really simple having an alias in Solidity. For that I would just suggest keccak and not ksha3.

However, only aliasing and renaming in Solidity might be even more misleading as one could expect the same hashing is used to create an address or ABI encoding.

[ethers]

It probably is unrealistic changing Keccak to SHA3 in all the above

Agree and that's a helpful list you provided. Before we consider outputting actual SHA-3 hashes, we need to start making the current naming more accurate, especially in the most visible code such as Solidity's sha3 function.

For Solidity, a new opcode isn't even needed: alias a new Solidity function to existing opcode, encourage use of new name by providing warnings in Solidity when sha3 is used, then possibly eventually compiler errors when sha3 is used in new Solidity code. Below the surface, the same existing opcode is used by Solidity.

[axic]

@ethers I guess an important question is: should SHA3 be available or only Keccak. If only the latter, a simple renaming in the EVM docs without changing the opcode number and aliasing and/or renaming in solc would be sufficient.

To clarify my first comment, by the new opcode I've meant a new opcode for SHA3 and renaming the current to Keccak; eventually supporting both in Solidity.

[ethers]

@axic We agree on "a simple renaming in the EVM docs without changing the opcode number and aliasing and/or renaming in solc" so I filed it for Solidity. The question you ask is indeed important, but I think sooner action is better esp on Solidity front since contracts are quite visible.

[Souptacular]

Just to demonstrate the difference, I posted a detailed comparison on this link and on this Ethereum StackExchange answer.

[emansipater]

Just adding my voice to reinforce how important this is. I have had to patiently explain this issue to just about every competent programmer I've ever introduced to Ethereum, and the problem is just growing over time. Actually swapping things out would be a significant challenge, but simply renaming is entirely doable and the sooner we do it the better. The longer we wait the more code, libraries, and implementations will have to be updated because they have continued to re-use the same naming. Let's create aliasing options to Keccak256 as soon soon as possible so that at least those cognisant of the issue can start tackling the issue in our own code and stop furthering the problem. And to be honest, if there's anyone using these functions who doesn't know on sight what Keccak256 is doing here when they see it, they need to learn that ASAP anyways, before they make a critical mistake because of it.

[axic]

Update: Solidity now has an alias named keccak256(). (The opcode is still called SHA3.)