• TehPers@beehaw.org
    link
    fedilink
    English
    arrow-up
    10
    ·
    1 year ago

    The only modern language that gets it right is Swift:

    print("🤦🏼‍♂️".count)
    // => 1
    

    Minor, but I’m not sure this is as unambiguous as the article claims. It’s true that for someone “that isn’t burdened with computer internals” that this is the most obvious “length” of the string, but programmers are by definition burdened with computer internals. That’s not to say the length shouldn’t be 1 though, it’s more that the “length” field/property has a terrible name, and asking for the length of a string is a very ambiguous question to begin with.

    Instead, I think a better solution is to be clear what length you’re actually referring to. For example, with Rust, the .len() method documents itself as the number of bytes in the string and warns that it may not be what you’re interested in. Similarly, .chars() clarifies that it iterates over Unicode Scalar Values, and not grapheme clusters (and that grapheme clusters are unfortunately not handled by the standard library).

    For most high level applications, I think you generally do want to work with grapheme clusters, and what Swift does makes sense (assuming you can also iterate over the individual bytes somehow for low level operations). As long as it is clearly documented what your “length” refers to, and assuming the other lengths can be calculated, I think any reasonably useful length is valid.

    The article they link in that section does cover a lot of the nuances between them, and is a great read for more discussion around what the length should be.

    Edit: I should also add that Korean, for example, adds some additional complexity to it. For example, what’s the string length of 각? Is it 1, because it visually consumes a single “space”? Or is it 3 because it’s 3 letters (ㄱ, ㅏ, ㄱ)? Swift says the length is 1.

    • neutron
      link
      fedilink
      arrow-up
      3
      ·
      1 year ago

      If we’re being really pedantic, the last part in Korean is counted with different units:

      • 각 as precomposed character: 1자 (unit ja for CJK characters)
      • 각 (ㄱㅏㄱ) as decomposable components: 3자모 (unit jamo for Hangul components)

      So we could have separate implementations of length() where we count such cases with different criteria… But I wouldn’t expect non-speakers of Korean know all of this.

      Plus, what about Chinese characters? Are we supposed to count 人 as one but 仁 as one (character) or two (radicals)? It gets only more complicated.