Adjust casing of namof operator

[petrroll]

Nameof operator currently translates to a string that's exactly the same casing as the symbol it represents. This limits it's usability in cases where the name of the symbol is required, but slightly transformed. For example from PascalCase to camelCase in case of JSON properties or many telemetry scenarios.

My proposal is to extend nameof(<symbol>) to something akin to nameof(<symmbol>, <oneOfClosedSetOfCasingOptions>).

var myNewFields;

Console.WriteLine(nameof(myNewFields, Casing.LowerCase));  // output: mynewfields
Console.WriteLine(nameof(myNewFields, Casing.PascalCase));  // output: MyNewFields
Console.WriteLine(nameof(myNewFields, Casing.CamelCase));  // output: myNewFields
Console.WriteLine(nameof(myNewFields));  // output: myNewFields

The set of casings should be closed and there should not be support for arbitrary transformations (it'd be hard/impossible to lower anyway without runtime impact) to limit the risks of this feature.

The set of casings could be somewhat similar to: https://learn.microsoft.com/en-us/dotnet/api/system.text.json.jsonnamingpolicy?view=net-9.0 + PascalCase and lower and uppercase.

Useful for:

• I want to enrich telemetry with fields of some class as tags (string key, string/object value) but telemetry is lowercase or camelCase: https://opentelemetry.io/docs/specs/semconv/general/naming/
• I want to have serialization fields follow some format / use different fields name for a field (ie can't just use name conversion strategy for the whole serializer) but the target uses different casing. s.a. camelCase so I can't use nameof in [JsonPropertyName]

https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/operators/nameof

[petrroll]

I understand that making e.g. lowercase thing pascalcase might be too complex / undesirable. As such, maybe just limit the feature to simplifying casing and never adding diffCasing where it wasn't would be best. It might be tricky to communicate however.

[petrroll]

Might be interesting to combine with #701

[yaakov-h]

This limits it's usability in cases where the name of the symbol is required, but slightly transformed.

That's not its name then, that's a different name.

[petrroll]

Yeah, but when you change the name you still want to change all locations most likely.
It's transformed name.

[huoyaoyuan]

C# language doesn't have the concept of character casing at all. For record parameters where doesn't matches the traditional naming convention, it still uses the originally specified string as constructor parameter name.

As a speaker of language without casing concept, I strongly disagree to bake the concept of casing into programming language.