and also i just realized that ?. is not called the elvis operator, the elvis operator is ?:

tell me if that works

nice! i was looking for that

🙂 happy to help

would it be theoretically possible to support type inference on arguments that have a default value?

it seems like that should be a thing considering types are inferred like everywhere else

IMO That would probably put some stress in the compiler since the value of default args can potentially be a function or value that also may use inference. It could also promote bad practices since public APIs should also always be properly typed. Types serve as documentation. In Scala most people started using inference everywhere and then I think most of the community agrees it was a bad idea and that the more typed the better. Not sure how that will play out in Kotlin land and I'm assuming the lang style guidelines will touch on this subject. Maybe someone from Jetbrains may weight in as to what their recommendation is in Kotlin.

Is there any word on if and when Kotlin can be used to build Cocoa iOS apps?

Hello everyone! I have just joined here 🙂

Hey guys, I've got an application that needs keep a map of about 100,000 entries. The key can be derrived from a decimal vector of about 100 columns. The value is also a decimal vector of about 100 columns. Some domain rules assert for use that the key vectors will always be unique. edit: I want to do this entirely in memory because I'm lazy.

so a sample might look like this:

<1.0, 2.0, ... 99.0, 100.0>    ->    < 0.1, 0.2, ..., 0.99, 1.00>
<1.1, 2.1, ... 99.1, 100.1>    ->    < 0.11, 0.21, ... 0.991, 1.01>

because

contentHashCode

on

DoubleArray

gives me back a 32 bit value, i dont have any gaurentees that its hash code wont have a collision, so i figured I might write my own hash implementation that uses a 128 bit hash

can I be certain this wont produce collisions?

Given that my input vectors are always unique, can i be confident that this will never produce a collision?

Chance of them colliding seems very slim, but why not run a batch of tests on a huge set to make sure?

a batch of tests on a huge set

hmm.

as in why dont I plug in some kind of data generator

I've written a handful of tiny static tests to at last do sanity checks

but some kind of load testing, that just iterates over a few million doubles, that kind of hting?

that seems like a pretty good idea,

Yeah, that's what I did when I wanted to test out unique ID's once. Even put it in impossible scenarios for added relief.

the problem is this test might ebb at the upper limits of the JVM's memory space. I wonder if that kind of GC thrashing would give JUnit a hard time?

I suppose I can find out

Give it a shot

I remember reading somewhere there's something like

val (key, val) = something something

What is that called?

It's called Destructing Declarations https://kotlinlang.org/docs/reference/multi-declarations.html

@groostav why not use the data itself as a key? Any hash will always have collisions (by definition) you want a 1-1 mapping. Also what you wrote is not a proper uuid so better not call it that.

So something I am struggling with, I am trying to build a key value map with mixed value types A value can be Int, String, Boolean, List<Value>, Set<Value>, or any random class.

Im trying to do this witout making it annoying for the user or reverting to Any and erasing the type info

@jkbbwr casting to any won't erase type info