Is it possible to reprioritize case var to be the same as default

[Thaina]

Since there is not possible to match anything below case var x I think it would be more natural to allow case var x in any order and treat it as equivalence to default. Actually the case var is more like mutually exclusive with default

I would like to propose that compiler should be like this

    void DoSomething(object obj)
    {
        switch(obj)
        {
            case var o: // should be treated the same as default. No matter where this case was placed it will be the last branch
                Console.WriteLine("obj is : " + o?.GetType() ?? "null");
                break;
            case int i: // should allow here instead of unreachable error
                Console.WriteLine("obj is int : " + i);
                break;
            default: // should error unreachable with this instead
                Console.WriteLine("Why can be here?");
        }
    }

The error logging logic could be more simplified in switch case with this

[CyrusNajmabadi]

No matter where this case was placed it will be the last branch

I really don't like that. 'cases' have never had that behavior. And it's been extremely easy to explain what happens with pattern cases that they simply match in order, and first wins.

[sharwell]

Not sure if there is already a warning if other cases follow case var x (or case _), but seems like that could be a reasonable solution.

[Richiban]

Yes you actually get a compiler error for provably unreachable cases: https://sharplab.io/#v2:EYLgxg9gTgpgtADwGwBYA0AXEUCuA7NAExAGoAfAAQCYBGAWACgKBmAAmtYGFWBvR1geypVe/QeIjAAVjDAZWkqawC8rAAwBuMeIHadAZwDuASwxgAFgApFASj3i+DHc9ZgAhvpisAbm6gKQexdxChoATksAIkVWY31WSNYSBQB+ADoAcRgMABUATwAHGEsbVhSUhLwcABtqyJstJ2CXYFg3AGtG5sF3T1i8eWNApu6Q8KiYuP6sBKTYhqDu1pgOrtHCGAAzNxqsRebQiMiAVTw2izdgaph6tecAXz1HhnugA===

There's no actual problem being solved in this issue, just the author really really wants to put the "default" case first in their switch.

[Richiban]

Why does var x get special-cased? As far as I'm concerned it's just a pattern and should behave the same as all the others: where a match is attempted in the order they are written.

Is it just var x? What about {} x?

[Thaina]

Because var x has highest coverage. It include null

It was the only case that will always make default become unreachable in control flow analysis. I cannot see anyway that these 2 cases could be mutually exist in the same switch

[HaloFour]

Not to mention case var can be combined with case guards.

[Thaina]

For those I think any case guard will be using the same current logic

Only the pure case var would be treated as default

Also because it was the only case that will always make default become unreachable in control flow analysis

[CyrusNajmabadi]

This gets confusing. As a pure-case-var would operate differently than a case var + with. This doesn't seem beneficial. If you want case-var to be handled last, list it last.

[HaloFour]

This doesn't seem beneficial. If you want case-var to be handled last, list it last.

Would be nice if that also applied to the evaluation order of patterns. :)

[Thaina]

I used to presenting alternative that there should be a way to define variable with default such as default obj: or default as obj: or default var obj: such that we can still get the variable from default case

What I want is that I can do the same logic but order it first because it was the smallest logic. Other logic would be longer below in switch

I was trying to propose this instead because default var has exactly the same use case as existing case var

[Thaina]

Pattern cases very intentionally do not do that

I mean, how is this possible?

        object obj = 0;
        switch(obj)
        {
            default:
                break;

            case int i:
                break;

            case string s:
                break;
        }

[HaloFour]

I really dislike that idea, especially since it would mean something very different with or without case guards. case var x should not behave differently than case var x when true.

[Richiban]

Given that order is significant in switch statements/expressions I really don't think the compiler should be silently reordering your cases.

default is an exception to this because it is absolutely 100% explicit that it should only be hit if all other matches fail.

[Thaina]

@Richiban I understand you reasoning. But then is it just familiarity?

For me I feel case var is special because its existence will precede default and make it become unreachable control flow

I agree that it might not be explicit enough. But there was no other choices since there is no expression that can attach to default to get a variable

Or is it?

[CyrusNajmabadi]

I mean, how is this possible?

'default' is not a pattern.