fix truncation of token flags - #4727

-Werror?
Edit: I guess you mean there was a warning before?

That makes no sense. mFlags is a uint64_t. Are you sure you have a clean checkout? What is your platform and compiler?

Visual Studio gave 2 warnings for the size mismatch before this patch and none after.

The warning I posted is currently disabled. Along with the other conversion warnings. Fixes for those are not wanted.

To be fair the signed/unsigned warnings were/are really buggy which makes things messy. But the implicit shortening/truncation warnings are safe.

Where is the truncation with this patch?

    uint64_t flags() const {
        return mFlags;
    }
    void flags(const uint64_t flags_) {
        mFlags = flags_;
    }
    enum : uint64_t {
        ...
        fIsInline               = (1ULL << 32), // Is this a inline type
        fIsTemplate             = (1ULL << 33),
        fIsSimplifedScope       = (1ULL << 34), // scope added when simplifying e.g. if (int i = ...; ...)
        fIsRemovedVoidParameter = (1ULL << 35), // A void function parameter has been removed
        fIsIncompleteConstant   = (1ULL << 36),
        fIsRestrict             = (1ULL << 37), // Is this a restrict pointer type
        fIsSimplifiedTypedef    = (1ULL << 38),
        fIsFinalType            = (1ULL << 39), // Is this a type with final specifier
    };
    uint64_t mFlags;

Changing the argument and return type back to unsigned int is obviously wrong because you are throwing away bits that are being used. This even fixes a bailout because fIsSimplifiedTypedef was being discarded. This is fixing a real bug that also happens to be a real and valid warning. I don't understand what the problem is.

Ah, now I get it. The warning is with the code before your patch.

See also cppcheck-opensource/simplecpp#270.

I'm not a developer anymore so can someone else merge this?

@IOBYTE good to see you hope you are well!

I'm doing good. Hope you and your family are too. Retirement is great but getting old sucks.