Proposal: null-conditional call

[dmitriyse]

Func<int, string> myFunc = null;
if (myFunc?(10) == null)
{
    Console.WriteLine("Every body happy now");
}

[HaloFour]

Func<int, string> myFunc = null;
if (myFunc?.Invoke(10) == null)
{
    Console.WriteLine("Every body happy now");
}

[dmitriyse]

Probably I need some sleep. :)
This proposal gives too few benefits, and we can close it. But if it have low cost, then it can be part of #35 improvement.

[alrz]

@dmitriyse

AFAIK This was considered in the design of C# 6.0 and discarded because of ambiguities with ternary.

[dmitriyse]

Then closing it.

[dmitriyse]

@alrz Please provide some case, where it gives ambiguity. If you remember.

[dmitriyse]

Probably it was declined due to cost/scheduling more than ambiguity?

[jnm2]

I remember the ambiguity too. Plus, writing out .Invoke makes the code easier to read imo.

[ufcpp]

@dmitriyse
It was due to ambiguity between null-conditional call f?(10) and conditional operator f ? (10) : 20.

[dmitriyse]

Good example, thanks.

[Mardoxx]

@ufcpp what ambiguity? f?(10) has no colon after it.

I see what you mean now...

Actually no, I don't see how this is reason for it to be rejected?

Unless this proposal would make `?` a unary operator, taking precedence over ternary, so the `f ? (10) : 20` would be interpreted as `x?(10) : 20` which is nonsense?

[CyrusNajmabadi]

The problem is that whne you see the ?( you don't have enough information to make the determination. The part that follows the ?( may be arbitrarily complex. This means you'd have to do an enormous complex scan (or an expensive speculative parse) to figure out what was going on. For example, if you had:

f?(a, b).ToString()

What is? Well, you still can't tell. Maybe it's your new nullable-call, or maybe it's a ternary that has a tuple on the 'true' side.

Given that you have full expression facilities available to you (i.e. 'a' and 'b' can be arbitrarily large), it's quite difficult to make a determination about what's going on.

Remember that resolving ambiguities doesn't mean just handling the simple case. It means handling all the cases :)

[Mardoxx]

Brilliant, thank you! I knew I was missing something :)

…

On 10 Nov 2017 8:39 pm, "CyrusNajmabadi" ***@***.***> wrote: The problem is that whne you see the ?( you don't have enough information to make the determination. The part that follows the ?( may be arbitrarily complex. This means you'd have to do an enormous complex scan (or an expensive speculative parse) to figure out what was going on. For example, if you had: f?(a, b).ToString() What is? Well, you still can't tell. Maybe it's your new nullable-call, or maybe it's a ternary that has a tuple on the 'true' side. Given that you have full expression facilities available to you (i.e. 'a' and 'b' can be arbitrarily large), it's quite difficult to make a determination about what's going on. Remember that resolving ambiguities doesn't mean just handling the simple case. It means handling all the cases :) — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#261 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAOoMY1oKYCyAzgbmGaGuyhAGD50CE4uks5s1LSWgaJpZM4MbPQV> .

[CyrusNajmabadi]

You're welcome :)