!! should be disallowed

[AlekseyTs]

        // PROTOTYPE(NullableReferenceTypes): Binder should report an error for `!!`.
        [Fact]
        public void SuppressNullableWarning_Multiple()
        {
            var source =
@"class C
{
    static void F(string? s)
    {
        G(default!!);
        G(s!!);
        G((s!)!);
    }
    static void G(string s)
    {
    }
}";
            var comp = CreateCompilation(
                new[] { source, NonNullTypesTrue, NonNullTypesAttributesDefinition },
                parseOptions: TestOptions.Regular8);
            comp.VerifyDiagnostics(/* ... */);
        }

[jcouv]

Relates to #25372 (warn for un-necessary !)

[gafter]

Was there an LDM decision to support this?

[jcouv]

I've updated the linked issue (warn for un-necessary !) after checking the LDM notes from last week. There was an explicit decision from LDM not to warn for such cases.
Regarding !!, there was no explicit decision from LDM yet.

[gafter]

I’m pretty sure that the decision not to warn included a decision not to error.

[jcouv]

The rationale for not warning (namely so that copy&paste doesn't produce annoying warnings) doesn't apply to '!!' which would be an error in any context.

[gafter]

Wouldn’t copy-and-paste (e.g. to inline a variable) now possibly cause an annoying error?

[jcouv]

Yes, that could happen.
Then, are you proposing to close this issue as "by design"?

[gafter]

Yes, I think it is permitted by design and not harmful.

[cston]

!! should be allowed. Need tests.
https://github.com/dotnet/csharplang/blob/master/meetings/2018/LDM-2018-08-20.md#remaining-questions-on-suppression-operator