Poor diagnostic message when using `?.[]`

[bwilkerson]

Create a file containing the following (invalid) code:

void f(String? s) {
  s?.[0];
}

The use of ?. before an index operator appears to no longer be supported, but the diagnostic message is very confusing. There is only one message generated (on the reference to s inside the body):

error: An expression whose value can be 'null' must be null-checked before it can be dereferenced. (unchecked_use_of_nullable_value at [nnbd_test_package] lib/test.dart:2)

The message should explain to users that they can't use ?. before [] to null-check s.

[FMorschel]

Today the message is a new one undefined_getter:

The getter '[' isn't defined for the type 'String'.
Try importing the library that defines '[', correcting the name to the name of an existing getter, or defining a getter or field named '['.

Although in this case probably s?[0]; or s![0]; should fix, if we had a getter/method we would see:

Maybe those fixes can be reviewed to show on this case and the ?. one could be transformed for this case as it is if you don't add the .:

---

I also noticed that this same fix doesn't show for function calls:

Maybe it could be reviewed for that as well (to add ?.call).

[lrhn]

Yeah, the new message isn't very good, since you can't define a getter named [.

Still poor diagnostic message.

The CFE says:

main.dart:2:15: Error: Expected an identifier, but got '['.
Try inserting an identifier before '['.
  print(args?.[1]);
              ^

(From DartPad.)
That's ... not bad. You might have meant to say args?.elements[1].
But it could also suggest args?[1], especially when args is a List?.

[FMorschel]

especially when args is a List?.

I would think the analyzer would know when there is an implementation for the [] (or []=) operator, or am I wrong here? That might be better than specifically List?.