if you only want equals() and hashCode(), the IDE can generate them for you

yeah I guess I will go that route

another option is to declare the primary constructor which takes the date as a date and the secondary constructor which takes the date as a string, parses it and calls the primary constructor with the parsed date

then you can keep this as a data class

hmm let me think about that one

I will go with a version with custom equals getHash, thanks for the alternatives

Does anyone know why methods like

IntArray.average()

allocate an iterator instead of using an index variable?

/**
 * Returns an average value of elements in the array.
 */
public fun IntArray.average(): Double {
    val iterator = iterator()
    var sum: Double = 0.0
    var count: Int = 0
    while (iterator.hasNext()) {
        sum += iterator.next()
        count += 1
    }
    return if (count == 0) 0.0 else sum / count
}

because they are generated from a template that is also used for other implementations of average(), and it’s not specialized for arrays

I filed an issue to fix this: https://youtrack.jetbrains.com/issue/KT-10579

Ah, that explains it, thanks simple smile

@yole: kudos on chapter 6 of "Kotlin in Action" very well written and easy to digest.

@mitchwongho: use case is for a sealed class hierarchy where all classes are immutable value types

e.g. like an algebraic data type in functional languages

You can implement it now, except you have to write all the functionality of a data class by hand (equals, hashCode, copy, component1, component2, etc. etc.)

I’d much prefer to write something like:

sealed data class Article(val title: String) {
    class RejectedArticle(title: String, val reason: String): Article(title)
    class AcceptedArticle(title: String, val doi: DOI): Article(title)
}

there might be information in Google for sealed classes as well. As on stack overflow, is good if people research first, ask second.

@natpryce: ran into this a lot recently. opted to just change my model which was disappointing but not unreasonable

<!everyone> who was having problems with the Kobalt build of u2020: it’s now fixed, pull (https://github.com/cbeust/u2020) What was going on:

Build.kt

captures a timestamp of the build so it can display it in the app. This is cute but it defeats incremental compilation (since it changes all the time, the build rebuilds everything all the time), so I temporarily replaced that string with “” but u2020 then blows up because it tries to parse this as a date. So I restored the build timestamp for now.

@cedric: wrong channel? #C0BRKJ9K9

oh crap, sorry everyone

doh

maybe if you delete it quickly, it doesn’t notify everyone?

womp woooomp

puts on dunce cap

feels really stupid

starts ranty diatribe about @ everyone

reminds @User to take it to #meta

😄

😄