Any reason to use the `Cids.sol` library?

[Stebalien]

As far as I can tell, this project could just use raw hashes. I ask because this library:

1. Does no validation.
2. Doesn't actually support arbitrary CIDs. It simply assumes that every CID is a prefix followed by a 32 byte digest.

IMO, it would be better to just use raw hashes.

[ZenGround0]

Does no validation.

By this do you mean validate that the prefix is particular mh / codec ? Or that mh length is correct?

[Stebalien]

It validates that the CID is at least 32 bytes long, that's it.

• It doesn't even look at the codec, length, or multihash-code.
• It doesn't support or even check if the CID has an unexpected digest length (CIDs support more than 32-byte digests).

From what I can tell, the "CIDs" here are all piece CIDs. I'd just store the digests. Saves space/gas anyways!

If we absolutely need to store them as CIDs, we should change this library to be PieceCid and have it check that the prefix matches a specific prefix and that the CID is of the expected length.

[Stebalien]

Alternatively, if you're worried about changing the state (even though you'd definitely save gas), change it to a generic slice manipulation library for adding/removing prefixes. That way, at the very least, nobody will be confused and think that this library actually handles CIDs in a sane way.