COMP: suppress Clang -Wduplicate-enum

[seanm]

No description provided.

[dzenanz]

Interesting style change by clang-format.

[seanm]

Interesting style change by clang-format.

Mmmm, which style change? I added more #if branches, so things got reindented...

[dzenanz]

I meant what clang-format wants by its complaint in pre-commit check.

[seanm]

I meant what clang-format wants by its complaint in pre-commit check.

Oh! fugly. :) Odd, because I ran clang-format -i locally before committing. I have version 20.1.6 though, maybe different.

[N-Dekker]

@seanm Thank you very much for addressing these warnings! I wasn't aware of them, when I proposed those enum's (PR #5410).

By the way, it appears that an enumerator like INT8 = CHAR would pass without a warning, even without your pull request:

ITK/Modules/Core/Common/include/itkCommonEnums.h

Line 92 in ab87323

INT8 = CHAR,

However, something like INT32 = std::is_same_v<int32_t, long> ? LONG : INT does indeed trigger that warning:

ITK/Modules/Core/Common/include/itkCommonEnums.h

Line 96 in ab87323

INT32 = std::is_same_v<int32_t, long> ? LONG : INT,

As I observed at https://godbolt.org/z/bPTzc68YK using clang 20.1.0 But unfortunately the std::is_same_v is necessary here anyway 🤷 So disabling them appears to be the best solution 👍

[N-Dekker]

I see that the if __has_warning is necessary to avoid another warning (warning: unknown warning group '-W...', ignored [-Wunknown-warning-option]). Unfortunately it's a bit cumbersome to have 5 lines of boilerplate code to define such a macro (ITK_CLANG_SUPPRESS_Wduplicate_enum), for just one warning. Would it be possible to make macro definitions like this more concise? And would it be possible to just specify the string "-Wduplicate-enum" once, instead of twice? I'm just asking. 🤷 No show stopper.

[seanm]

I meant what clang-format wants by its complaint in pre-commit check.

maybe I've fixed it with my change to the clang-format config file...

[N-Dekker]

Thanks @seanm

If you ever find a way to define macro's like ITK_CLANG_SUPPRESS_Wduplicate_enum in a more concise manner, please let us know. Approved anyway 👍

[thewtex]

Thanks @seanm 🙏

[N-Dekker]

By the way, it appears that an enumerator like INT8 = CHAR would pass without a warning, even without your pull request

To further elaborate:

  enum class MyEnum
  {
    UNKNOWN,
    CHAR,
    INT8 = CHAR,  // Intended duplicate value, no problem, no warning!
    WONDERFUL = 42,  // OK, new enum value
    DUPLICATE_ZERO = 0  // <== Warning! Duplicate value!
  };

See https://godbolt.org/z/78r883836