As of today, the registry assumes that there is only one event per signature but in reality different, binary incompatible events can have similar signature because of indexed values.
ERC20
event Transfer(address indexed from, address indexed to, uint256 value);
This way, it's hard to write code that uses registry to parse raw logs without additional complex logic to handle such collisions.
What if registry would return a list of possible event representations? This way, the client can just go over the list trying to decode each possible combination.
The possible issue that I can see is that the registry would be spammed by all possible combinations of indexed/unindexed fields, most of them being useless. For transfer, you have 2^3=8 possible combinations, and some of them are indistinguishable because it does not matter to binary representation which argument is indexed. I think sorting variants by "popularity" can resolve some issues.
What do you think about this problem, and can it be realistically solved in registry?
As of today, the registry assumes that there is only one event per signature but in reality different, binary incompatible events can have similar signature because of indexed values.
Notable example - transfer event https://www.4byte.directory/event-signatures/?bytes_signature=0xddf252ad1be2c89b69c2b068fc378daa952ba7f163c4a11628f55a4df523b Can be of two different flavors
event Transfer(address indexed from, address indexed to, uint256 indexed id);event Transfer(address indexed from, address indexed to, uint256 value);This way, it's hard to write code that uses registry to parse raw logs without additional complex logic to handle such collisions.
What if registry would return a list of possible event representations? This way, the client can just go over the list trying to decode each possible combination.
The possible issue that I can see is that the registry would be spammed by all possible combinations of indexed/unindexed fields, most of them being useless. For transfer, you have 2^3=8 possible combinations, and some of them are indistinguishable because it does not matter to binary representation which argument is indexed. I think sorting variants by "popularity" can resolve some issues.
What do you think about this problem, and can it be realistically solved in registry?