Why does codecvt_utf8_utf16::do_max_length() return 6 in MSVC2017?

119 views Asked by At

Why does Visual C++ 2017 implementation of std::codecvt_utf8_utf16<>::do_max_length() returns 6 (assuming codecvt_mode is 0)?

        // CLASS codecvt_utf8_utf16
template<class _Elem,
    unsigned long _Mymax = 0x10ffff,
    codecvt_mode _Mymode = (codecvt_mode)0>
    class codecvt_utf8_utf16
    : public codecvt<_Elem, char, _Statype>
    {   // facet for converting between UTF-16 _Elem and UTF-8 byte sequences

    // ...

    virtual int __CLR_OR_THIS_CALL do_max_length() const _THROW0()
        {   // return maximum length required for a conversion
        return ((_Mymode & consume_header) != 0 ? 9 // header + max input
            : (_Mymode & generate_header) != 0 ? 7  // header + max output
            : 6);   // 6-byte max input sequence, no 3-byte header
        }

Shouldn't it be 4 instead (to encode any Unicode code point in UTF-8 at most 4 bytes are required)?

0

There are 0 answers