Minor Inconsistency String#bytes vs Bytes type

[lribeiro]

typeof("324".bytes) -> Array(UInt8)
typeof(Bytes) -> Slice(UInt8)

Shouldn't the String#bytes method, return a Slice(UInt8) instead of an Array(UInt8) ?

[asterite]

An array is usually more useful as it's mutable and it's a copy of the underlying string bytes. If you want the bytes as a slice for performance reasons you can use to_slice or to_unsafe.

[asterite]

I mean, if somebody eventually introduces, say, a Downcase type, we then have String#downcase and it would be inconsistent because it returns a String? I don't think so.

Also the docs say: "Returns this string's bytes as an Array(UInt8)"

[Sija]

@asterite There's also Slice#to_a to easily do the opposite :)

[asterite]

I don't understand, we also have #chars and #codepoints. Should these return Slice instead of Array? Array is the preferred data type to return because it's the most flexible and useful, and it's the one that is commonly returned in API methods.

[lribeiro]

When working with Bytes it mentally trips me to use String#bytes because of the name similarity.

@asterite Ruby for instance does return an Array. But it as no Slice equivalent on std lib (to my best knowledge, didn't check).

Why is Bytes an alias for Slice(UInt8) and not Array(UInt8)?

I don't have a strong preference either way, I just think it would be better if they are consistent.
It's not a big deal :)

[asterite]

I was kind of against the Bytes alias... maybe this explains why :-)

Slice(UInt8) is the most common slice type and it's all over the place, so we thought Bytes would be a good alias for it. Array(UInt8) it not a very common type at all. In fact, maybe we could remove String#bytes for that matter...

[lribeiro]

I found Bytes by accident reading the source code and started using it on my code, I thought it was kind of cool and it does sound better than Slice(UInt8).

If it is renamed to String#to_a that would open up space for String#bytes returning Bytes :)

[kostya]

thats why i vote for name Binary :)

[lbguilherme]

A better name for this could be "aaa".utf8. The idea is that while all you have is a String, the fact that it is stored as an UTF8 sequence of bytes is just an implementation detail that is not disclosed. But this is the method that exposes this binary representation.

[asterite]

I'm closing this, this is a minor consistency, and I'm sure there are many more bigger inconsistencies throughout the language. It's really impossible for a language to be 100% consistent. Bytes (or Slice(UInt8) is mostly used in low-level code to read/write from an IO. Array is the preferred container type, so it's OK for String#bytes to return Array(UInt8).

[j8r]

To clarify the method, it could be String#bytes_array : Array(UInt8).