feat: add base256emoji

[rvagg]

Ref: multiformats/multibase#88

Too much fun, I couldn't help myself. Not intended to be performant of course.

/cc @Jorropo

[Jorropo]

I don't understand all the details of the implementation.

It seems like you made base256emoji a special cases which a bunch of adhoc rules.
I would try to avoid that since this is an increased load on the future maintainers (including future you).

[rvagg]

There's two special cases in here:

1. The this.baseDecode(text.slice(1)) normal-path which doesn't work for unicode strings, we just don't have a nice, efficient way to slice off the first multibyte unicode character in JS. So I'm special-casing it here so I can futz with Array.from().slice().join(), which is much more inefficient; but we're accepting that for this case. So without an efficient char slicing mechanism I'm fine with this. 🤷
2. In the tests there's a special case because I'm not including base256emoji in the 'multiformats/basics' module that's a kitchen-sink export, so I have to add it back in where it's not there. That's maybe a line-ball call, perhaps it could be included in basics and that special-case would go away. Again, 🤷, maybe @Gozala has opinions on these though.

[Jorropo]

we just don't have a nice, efficient way to slice off the first multibyte unicode character in JS

If JS doesn't have utf8.DecodeRuneInString then I guess it make sense. This path could be generalised if an other unicode base is made.

which is much more inefficient; but we're accepting that for this case

👍 this is clearly a negative priority item, being slow is fine

[rvagg]

If JS doesn't have utf8.DecodeRuneInString

Well .. Array.from() will do a nice job, but the contract for decoders is that they currently accept a string with the first character lopped off. So I could use Array.from() on all of them and pass in an array of characters, and that would be really nice, but then this is a breaking change because we allow arbitrary external implementations of Multibase (I don't know if any exist, probably not, but the contract exists).

We could do that in a future release though, it would be nicer than a slice(1), you could then just iterate over a proper array rather than a string.

[Gozala]

Well .. Array.from() will do a nice job, but the contract for decoders is that they currently accept a string with the first character lopped off. So I could use Array.from() on all of them and pass in an array of characters, and that would be really nice, but then this is a breaking change because we allow arbitrary external implementations of Multibase (I don't know if any exist, probably not, but the contract exists).

I think contract is that they pass in string without multicodec tag, so far that meant first character, but if tag is first emoji in the sequence it would mean snipping it off.

[rvagg]

Updated with s/☉/🐉 from the latest spec change for this

[github-actions]

🎉 This PR is included in version 9.7.0 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀