[Proposal] Shorter Alternative To Ternary Operator: Secondary Operator

[jwbats]

We can write if statements, to conditionally assign values, like such:

if (success)
{
	myObject.Value = "New Value";
}

But some of us would rather eliminate control flow, so we write it using the ternary operator:

myObject.Value = (success) ? "New Value" : myObject.Value;

The problem is that you are forced to redundantly write "myObject.Value", to ensure this property keeps its old value, in case of !success.

It would be nice if you could write it like this:

myObject.Value = (success) ? "New Value";

Granted, this could also be done using an extension method:

public static class ConditionalExtensions
{
    public static T ApplyIf<T>(this T original, bool condition, T newValue)
    {
        return condition ? newValue : original;
    }
}

But the resulting call still suffers from redundant property usage:

myObject.Value = myObject.Value.ApplyIf(success, "new value");`

There is currently no clean, shorthand way of writing a conditional assignment in C#.

It would be nice to have it, IMO.

I guess it would have to be called the "secondary operator", then?

[HaloFour]

See: #197

[jwbats]

It is indeed pretty much the same request, although the author doesn't provide an example of how he'd like to write it.

It's from 2017 and it's still not in the language.

I guess chances of this one being accepted are slim to none.

But why not... would it be that hard to implement?

[HaloFour]

would it be that hard to implement?

It is ambiguous, as expressions containing ternary operators can be nested and without a 1:1 relationship between ? and : it becomes impossible for the compiler to determine which is associated with which. Even if that weren't the case, or the compiler could figure out some kind of rules to disambiguate, it's still very little reward for the amount of effort that it would require as a language change.

[jwbats]

Good point.

You'd have to come up with another (combination of) special character(s) to serve as an operator, specifically for this use case.

I'm not sure which one(s) would serve best.

Might still be worth it, though.

[TahirAhmadov]

How is this:

myObject.Value = (success) ? "New Value" : myObject.Value;

Better than this?

if (success) myObject.Value = "New Value";

[jwbats]

My proposal is for a secondary operator, that will let me assign conditionally, without having to resort to an if-statement, and without having to resort to the redundancy of the ternary operator.

My proposal above accomplishes this in a clean, shorthand, non-ambiguous way.

Btw, are you part of the C# implementation team?

[HaloFour]

Btw, are you part of the C# implementation team?

Nope. But I have been following these repositories long enough to know how the team feels about shifting around conditionals to affect statements for stylistic purposes. The if-statement works perfectly well and is quite concise, there's no reason to add another flavor of it.

[jwbats]

That's your opinion.

Thanks for your input, though.

[TahirAhmadov]

There is a fundamental problem with your proposal.

If you don't like imperative control flow, like an if condition, then adding another operator to achieve it is not really justified - C# (and most other languages) already have if statement. In general, functional programming does not preclude conditional execution flows; without them, it becomes more like an Excel spreadsheet with formulas - just a giant complex calculator.

If you don't like the if statement itself, it becomes a subjective preference of yours. As previously mentioned, it's probably better to have the if in the beginning of the statement.

[jwbats]

All opiniated.

My proposed operator makes the statement equally long, so we can't say it's shorter or simpler.

It is, however, clearer, because the assigned-to variable goes to the left.

if (success) a = true;

a = (success) ?; true;

It's unfair to state that a proposal has a fundamental problem, when you merely don't like it, because of your own subjective preference.

I have a certain methodology of programming, in which this proposed operator would look really good.

I had this idea in my head for a long time, and I thought I'd just propose it.

if it doesn't happen, no problem. I wrote it down, so now it's off my mind.

FOr this reason, I'm not really interested in discussing it further.

[KennethHoff]

Personally I think some syntax like this would be very useful as in combination with the spread element (..e).

bool greetTheWorld = ..;

ReadOnlySpan<string> greeting = [
  "Hallo",
  greetTheWorld && "World"
];

ReadOnlySpan<string> thanks = [ "Thanks", "for", "all", "the", "fish" ];

ReadOnlySpan<string> finalString = [.. greeting, ..(isDolphin && thanks)]

[TahirAhmadov]

I think what you are asking for is conditionally adding elements to the collection, and I think it will be supported with ["Hallo", ..(greetTheWorld ? ["World"] : []) ]

[RikkiGibson]

When it comes to "conditionally adding collection elements", I could see us doing something special, if there is sufficient feedback that [a, b, .. cond ? elems : []] is just not good enough.

I do not think there is any need to parenthesize the conditional-expression here, FWIW. The .. is not part of the expression here. It's just a special token when used in a collection element position. So, there is no risk of it being interpreted as (.. greetTheWorld) ? ["World"] : [], for example.

[jnm2]

["Hallo", ..(greetTheWorld ? new[] { "World" } : []) ] works today.

[NN---]

So basically you want “if” as expression rather than a statement.
Some languages have it and also with “if” syntax.