Allow equality checks between string types #1507
Conversation
|
@jacqueswww updated 👍😄 |
| typ=BaseType('bytes32'), | ||
| pos=getpos(expr), | ||
| ) | ||
| elif sub.location == "storage": |
There was a problem hiding this comment.
@jakerockland I just read this now, after having merged it. Will this not only hash the first 32bytes?
There was a problem hiding this comment.
@jacqueswww hrm yeah this is an oversight on my part too 😓
Given that this is the same code generation that is (and was) being used for the built-in keccak256 function as well, I think the issue here is a bit broader in scope than the string comparison stuff, in that for any hashes of any string literal/bytes array longer than 32 bytes, we are only hashing the first 32 bytes.
Given our definition of the built-in supporting string literals + byte arrays:
def sha3(a) -> b:
"""
:param a: value to hash
:type a: either str_literal, bytes, bytes32
:output b: bytes32
"""I think we may want to open another ticket specifying that we need to fix the keccak256 behavior all-around to hash all bytes + need to add tests around this both in the context of string comparison but also just using the built-in.
I will have some time this evening that I can get started with writing tests around this to verify the issue + can open up a ticket from there.
Thoughts? 😅
There was a problem hiding this comment.
Hmm just check again, pretty sure keccak256/sha3 has tests for longer than 32 byte strings. Maybe only storage values ?
There was a problem hiding this comment.
Hrm, yeah you're right we do have a couple of tests for that in tests/parser/functions/test_keccak256.py.
I need to do a bit more reading on the LLL here, but it does seem to me that if things currently work properly for hashing byte arrays longer than 32, that string comparison will work as well, as they use the same helper method?
Likewise, if there was an issue with the string comparison for strings/byte arrays longer than 32, it would make me think we need to add some more tests with the keccak256 built-in.
Does that make sense? @jacqueswww
There was a problem hiding this comment.
Either way, it probably would be good to add more nuanced tests to both functions/test_keccak256.py and types/test_string.py, which I'm happy to do this evening @jacqueswww — from there it will be a lot more clear what exactly is wrong 😅
There was a problem hiding this comment.
Yeah, clearly more tricky than one would initially think! :)
What I did
Implemented VIP #1355, allowing equality checks between string types.
How I did it
Check strings with length > 32 or unequal lengths by comparing
keccak256hash of each.How to verify it
Run the tests:
make test. More specifically:pytest -k test_stringandpytest -k test_bytesDescription for the changelog
Allow equality checks between string types
Cute Animal Picture