Improve Support for NonEmptyChunk #3374
Conversation
| case Chunk.Empty => Chunk.single(a) | ||
| case ne: NonEmptyChunk[A] => Chunk.concat(ne, Chunk.single(a)) | ||
| } | ||
| final def +[A1 >: A](a: A1): Chunk[A1] = |
There was a problem hiding this comment.
this could still return NonEmptyChunk
There was a problem hiding this comment.
We could but there is potentially a little extra overhead in creating the NonEmptyChunk which I thought you might not want if you were doing repeated appends and were doing high performance stuff where you didn't care about whether it was empty or not (e.g. you're just calling foreach on it). So right now you kind of stay in the Chunk world unless you explicitly go to the NonEmptyChunk world, and once you're there we preserve that information whenever possible. But we could definitely change as you suggest.
| * `NonEmptyChunk`. Operations on `NonEmptyChunk` which could potentially | ||
| * return an empty chunk will return a `Chunk` instead. | ||
| */ | ||
| final class NonEmptyChunk[+A] private (private val chunk: Chunk[A]) { self => |
There was a problem hiding this comment.
Could we make this a value class?
There was a problem hiding this comment.
Potentially. The issue there is that then it is the same class at runtime as Chunk so methods that have overloads for Chunk and NonEmptyChunk like foreach cause compiler errors. We could possibly resolve by adding a dummy parameter to them.
Resolves #3357.