@miha-x64 You can't pass block to block when it's crossinline

noinline

then 🙃

Then you should just remove inline from the function altogether 🙂

there is too little bytecode for creating new method 🙂

sometimes I use

inline

methods without lambdas

Actually, what would be easier and even sufficcient for probably a lot of use cases is to have very limited classes, which allow these kind of fancy stuff more easily. I'm thinking of C++ struct type here, without methods and other "fancy" class-stuff.

If you are still interested:

@User uploaded a file: Untitled

Something like that would be interesting, although it would be a more limited use case.

or another possibility: allow meta-programming, so we can just write functions that work on class declerations 😛

How do you iterate over a range of doubles? or even just from on double to another? I can’t get any of the range/foreach logic to work for doubles (while it works for integers)

Iterating on doubles is considered harmful, since it exposes you to a host of nasty rounding errors. One usually iterates on integers, then multiplies an integer by the corresponding double step.

this came up pretty recently, last couple weeks. search for “range float” in chat for the full discussion. but basically the ranges are checked at either end using equals, not greater-than or less-than.

clever solution

@elizarov While I understand there are rounding issues, I’m wondering if that is really an issue for most peoples ranges/for-loop needs, or if there is an argument to made for the traditional for loop syntax. I know this was quite a point of pain for me initially. I ultimately ended up doing exactly what you proposed, but it takes another mental step when processing/reading the code to decipher in my opinion.

Then again ranges being limited to integers is not uncommon. So I’m leaning towards “how can we make looping over float values more native/simple?”

Use a while loop

I find a lot of arithmetic loops more complicated in Kotlin than they would be in Java due to the loss of the 3-part for loop. Although I guess it might be possible to produce something similar now that I think about it.

@kennycason I don't know if it would be an issue for "most" ranges, but in the cases where it is problematic it will make for insidious bugs.

@Ruckus I have used a while loop as well, but it’s not as compact and nice as a 3 part for loop or range. 🙂

Since the workaround is pretty simple (and, imho, makes for cleaner looking code) it seems like a good idea to avoid the insidious bugs.

For example, if your step size is 1e-10, your loop will never finish if the terminal condition is above 1e7.

Personally I've found that when I'm trying to use floating values in loops there's one of two cases: 1. The floating values are consistantly spread, and translating to integer values usually makes the code clearer (though it sometimes requires cleaning up the logic too) 2. The values change by arbitrary amounts, and a while loop makes it clearer (and more deliberate)

Just my $0.02

@jo Would that be any different than iterating over doubles in a 3 part for loop?

I can echo Ruckus' sentiment.

@kennycason My point was more, 'do it with integers so you know your loop will terminate'.

Basically echoing Ruckus' sentiment.

Usually, if I'm doing ML or game dev stuff that has to step by a float amount, I'll have a 'stepSize' = 1.0e6, then will multiply by that inside the intereger loop.

Also, for what it’s worth, the coord system in my game is all floats