Android studio often suggests me that I can use stdlib operation i.e.

filter{}

,

map{}

over normal loop code. Yes, I like to use those. But peek into the code from IDE reveals that actually

filter{}

is not better than normal loop as it dynamically creates a new

ArrayList

object to hold filtered elements. So normal loop would slightly has better performance. My question is as IDE suggests me like this - Is the performance that significant (although I can’t feel any difference)? - Is there anyone already did performance test comparing between

filter{}

,

map{}

etc against normal way? - Should I always use lambda functions as also suggested by IDE? (I want to but from above, I didn’t feel safe)

haxpor: With your loop, do you filter it in place or something? Is it mutable?

I captured the screen for the code before I changed them all to normal loop. It looks like this https://pbs.twimg.com/media/DEqGVA3VwAA-I7a.jpg:large

Ok, I think would do something like this:

flyingSaucers.filter { !it.shouldBeRemoved }
             .flatMap { it.bullets }
             .filter { !it.shouldBeRemoved && player.contains(it.x, it.y) }
             .forEach {
                 it.shouldBeRemoved = true
                 player.hit()
                 Game.res.getSound("explode")?.play()
             }

Thanks, I understand it might be premature. But are you sure it’s in-place?

No, it is not in place. I was wondering what you did with your loops.

Ohh, I see. Yup.

You could convert it to a sequence with

.asSequence()

to make it lazy though, but I'm not really sure if it would make a difference here.

still

filter

will require to create ArrayList to hold result, but

asSequence()

is better too I think.

It could be that the JIT will optimize it away even. It shouldn't be a problem.

I totally didn’t think about

takeIf

or

takeUnless

, so I can combine

forEach

with

takeIf

without creating new object in

filter

before going into the loop.