Why should I write CLS compliant code?

Question

I've found a lot of pages about CLS compliance.

I've understood that CLS compliance:

• Is a way to guarantee different assembly compatibility.
• Is a way to declare the high security code

Many peolple write that "if you write code, you should write it CLS compliant." But as far I can read, there is no reason to use CLS compliance in generic software.

Am I right, or did I miss something?

Secularism

related #570952 – Righteous

Answer 1

If you write a library or framework it makes sense to ensure your library can be used from any CLR language.

Slipsheet

CIL (Common Intermediate Language) is not sufficient? – Secularism

afaik no, because CIL may contain features that are unusable from certain CLR languages. Eg. you have a method that takes UInt32 parameter, the CIL code is fine but the method cannot be invoked from a language that has no concept of unsigned. – Slipsheet

What happens is the language is "translated" in CIL (having the source code, of course)? In this scenario, it seems CLS compliance is required only with legacy assemblies, right? – Secularism

No, it is required if the DLL is to be used from any langauge that only supports the CLS SUBSET of the CLR functionality. – Rosaceous

@TonyD Common Language Specification – Slipsheet

Not to nitpick, @TonyD, but its probably easier to just google that than ask it in a comment on here. msdn.microsoft.com/en-us/library/… – Buchanan

Answer 2

CLS-compliance is particularly important if you're distributing libraries - specifically, writing CLS compliant guarantees that your libraries will be usable by all CLS-compliant languages.

For instance, Visual Basic is not case-sensitive, whereas C# is. One of the requirements of CLS compliance is that public (and protected) member names should not differ only by case, thus ensuring that your libraries can be used safely by Visual Basic code, or any other .NET language that doesn't differentiate based on case.

Wulf

There's also a rule that you can't start members with underscores. Which leaves me wondering how to name a protected variable that is wrapped by a public property... Can't use a lowercase version, can't use an underscore... what's the next ugly alternative? – Mcnully

Probably prefixing it with a p. Not pretty, but functional. While I prefer C# Properties over Java-style getters and setters, at least in Java the naming convention is easier. – Wulf

Somehow, when it's all said and done, java ends up being more concise. I think it's probably more the libraries than the language, though. – Mcnully

PowerShell is also case-insensitive, which has always left me wondering how it'd behave when a .NET class contains members differ only in case – Glissando

Answer 3

The answer is to allow maximum compatibility across .NET languages. CLS is the lingua franca that allows C# assemblies to work with F#, Iron Python, C++/CLI, VB.NET, Boo and all the other .NET languages. Step outside that boundary and your assembly may work correctly, but not necessarily.

Detonation

Answer 4

There may not be a specific reason to have your code be CLS compliant, but people are referring to it being a "best practice"--something that you should do because it's a good habit, rather than being measurably better for a particular scenario.

In other words, it's a good idea to make your code CLS compliant unless you have a reason not to.

Tacket

I find it very strange that many .NET libraries aren't CLS compliant. So pretty much any class that uses one of those components isn't compliant either. – Satire

@MrFox: Just using such a component is fine. Publicly exposing it makes your class non-compliant. – Cannelloni