IntelliJ suggests to convert `r1` to `r2`, but the...

# announcements

[Kroppeb]

IntelliJ suggests to convert r1 to r2, but these are not equivalent. (The inferred type isn't even the same) Seems to be an issue if an extension function on a nullable type gives a generic type with the input type as one of the type parameters and as input a nullable type with a nullable parameter

[streetsofboston]

The inferred types are different because of the nullability of T.p, correct?

[streetsofboston]

Maybe it suggests the conversion because a List<Int> is a sub-type of `List<Int?>`… just guessing here 🙂

[Frodrigues]

That's only a suggestion. But you really want to make a list with a null?

[Kroppeb]

If t = T(null), then r1 will return [null] but r2 would return [10]

[w_bianrytree]

Try add a ? at your capture function?

[Kroppeb]

If you mean making the type argument non nullable, no. That’s not the problem. The problem is IntelliJ giving a suggestion that doesn’t work.

[bodiam]

In this case it seems maybe better to create Youtrack issue in the IntelliJ project, since this looks like a bug.

[Alexey Belkov [JB]]

Yes, please report to http://kotl.in/issue with sample code to reproduce in text form. Thanks!

[Kroppeb]

https://youtrack.jetbrains.com/issue/KT-34813