Wrong folding of the ternary operator

[sabi0]

Or is it some special (Kotlin?) syntax when "Elvis" condition can be a part of a bigger expression?

[stoerr]

I too was puzzled by a strange folding related to the elvis operator. In Java and JSPs I stumbled over foldings of

Something something = res != null ? res.methodcall(whatever) : null;

to

val something = res?.methodcall(whatever) ?: null;

This is not quite correct. According to the Java elvis operator proposal A ?: B would unfold to A != null ? A : B , which does not fit that folding.

And the ?. operator, e.g. the save call operator in Kotlin, has A ?. B unfold to A != null ? A.B : null . This also doesn't fit the folding above.

I think the line should fold to

val something = res?.methodcall(whatever);

If you literally would unfold what it folds to now, you'd get

Something something = (res != null ? res.methodcall(whatever) : null) != null ? (res != null ? res.methodcall(whatever) : null) : null;

which makes no sense at all. 8-)

On second thought, this might be a different issue, but the original point also treats ?: wrongly. Actually, A ?: null is the same as A, so it might be sufficient to omit something like ?: null and use ?: only if there is something non-null after it.