Looking to exchange some ideas around a strange KM...

# multiplatform

[ade]

Looking to exchange some ideas around a strange KMM behaviour.
We had this function in a commonMain source for Android and iOS code;

private fun String.strip() = this
    .replace(oldValue = "https://", newValue = "", ignoreCase = true)
    .replace(oldValue = "http://", newValue = "", ignoreCase = true)
    .substringBefore(delimiter = "?")

which was called from the same file on a String. This has been working fine until recently. Now suddenly, the function does nothing ("https://" prefix is not removed from a string) on Android but works like before on iOS. I think it was after bumping to Kotlin 1.7.20, AGP 7.3 and compileSdk 33. What could be happening here?

Theory: Is something causing it to call through to the Java String.strip() function instead, which was added in Java 11? Does it have to do with desugaring?

[Matthias Geisler]

Oh good to know - did you report that? Does that occurs without KMP? What is with Kotlin 1.7.21?

[ade]

I have not reported this yet

[ade]

Do not yet know if it happens without KMP, or if there is a change with 1.7.21

[ade]

The behaviour is the same with Kotlin 1.7.21

[ade]

I've made a minimum reproducible example and it appears to be triggered by changing to compileSdk = 33 in the kmp module. However, in my repro code project I get a warning in the IDE to not use ".strip()" which I didnt get in my main project...

[ade]

I also get a warning on my definition of my extension function, which I didnt get in the other project either

[ade]

of course, none of those warnings help on old code that continues to compile when bumping from compileSdk = 32 to compileSdk = 33.

[ade]

This problem in itself is in fact not related to KMP/KMM, other than that behaviour is different when compiled via KMP to iOS

[ade]

It isnt caused by desugaring, by the way, Ive tried with that both on and off and the behavior is the same

[ade]

Ok so after some reading, this is by design: https://kotlinlang.org/docs/extensions.html#extensions-are-resolved-statically

"The member always wins"

And unfortunately for us, we used a too generic name for the function ("strip") and its been added somehow when using compileSdk = 33, so it wins.

In my opinion that should have caused a compile error.