Question

我正在使用一个包含c文件的C ++项目（c文件编译选项是/ TC），项目编译完全正确。

然后我需要使用一些Pinvokes，我需要使用/ clr编译项目但是我必须使用/ TP作为C文件，当我这样做时大喊：

未解析的外部符号dtoa_grisu3 但是这个函数是在.c文件下定义的，它只在用/ TP编译时才会大喊。

这是标题：

#pragma once

extern "C"
{

/// Converts the given double-precision floating point number to a string representation.
/** For most inputs, this string representation is the
    shortest such, which deserialized again, returns the same bit
    representation of the double.
    @param v The number to convert.
    @param dst [out] The double-precision floating point number will be written here
        as a null-terminated string. The conversion algorithm will write at most 25 bytes
        to this buffer. (null terminator is included in this count).
        The dst pointer may not be null.
    @return the number of characters written to dst, excluding the null terminator (which
        is always written) is returned here. */

int dtoa_grisu3(double v, char *dst);

}

#ifdef __cplusplus

#include <string>
std::string dtoa_grisu3_string(double v);

#endif

这是植入（在c文件中）：

int dtoa_grisu3(double v, char *dst)
{
    int d_exp, len, success, decimals, i;
    uint64_t u64 = CAST_U64(v);
    char *s2 = dst;
    assert(dst);

    // Prehandle NaNs
    if ((u64 << 1) > 0xFFE0000000000000ULL) return sprintf(dst, "NaN(%08X%08X)", (uint32_t)(u64 >> 32), (uint32_t)u64);
    // Prehandle negative values.
    if ((u64 & D64_SIGN) != 0) { *s2++ = '-'; v = -v; u64 ^= D64_SIGN; }
    // Prehandle zero.
    if (!u64) { *s2++ = '0'; *s2 = '\0'; return (int)(s2 - dst); }
    // Prehandle infinity.
    if (u64 == D64_EXP_MASK) { *s2++ = 'i'; *s2++ = 'n'; *s2++ = 'f'; *s2 = '\0'; return (int)(s2 - dst); }

    success = grisu3(v, s2, &len, &d_exp);
    // If grisu3 was not able to convert the number to a string, then use old sprintf (suboptimal).
    if (!success) return sprintf(s2, "%.17g", v) + (int)(s2 - dst);

    // We now have an integer string of form "151324135" and a base-10 exponent for that number.
    // Next, decide the best presentation for that string by whether to use a decimal point, or the scientific exponent notation 'e'.
    // We don't pick the absolute shortest representation, but pick a balance between readability and shortness, e.g.
    // 1.545056189557677e-308 could be represented in a shorter form
    // 1545056189557677e-323 but that would be somewhat unreadable.
    decimals = MIN(-d_exp, MAX(1, len-1));
    if (d_exp < 0 && len > 1) // Add decimal point?
    {
        for(i = 0; i < decimals; ++i) s2[len-i] = s2[len-i-1];
        s2[len++ - decimals] = '.';
        d_exp += decimals;
        // Need scientific notation as well?
        if (d_exp != 0) { s2[len++] = 'e'; len += i_to_str(d_exp, s2+len); }
    }
    else if (d_exp < 0 && d_exp >= -3) // Add decimal point for numbers of form 0.000x where it's shorter?
    {
        for(i = 0; i < len; ++i) s2[len-d_exp-1-i] = s2[len-i-1];
        s2[0] = '.';
        for(i = 1; i < -d_exp; ++i) s2[i] = '0';
        len += -d_exp;
    }
    // Add scientific notation?
    else if (d_exp < 0 || d_exp > 2) { s2[len++] = 'e'; len += i_to_str(d_exp, s2+len); }
    // Add zeroes instead of scientific notation?
    else if (d_exp > 0) { while(d_exp-- > 0) s2[len++] = '0'; }
    s2[len] = '\0'; // grisu3 doesn't null terminate, so ensure termination.
    return (int)(s2+len-dst);
}

如何使用/ clr编译项目并避免未解决的符号错误？

Answer 1

您需要使用extern "C"或包含相关的headerfile [preferred]以确保它具有相同的名称。该名称因C ++函数而受到损坏，但extern "C"的名称不会受损，因此包含标题的调用将生成＆＃34; unmangled＆＃34; name，函数定义将具有错位名称 - ＆gt;链接器无法找到它。

Answer 2

有什么问题？

dtoa_grisu3中的extern C具有 C链接

C文件中的dtoa_grisu3具有 C ++链接

它们是不同的符号，这意味着C文件中的定义不适用于头文件中的声明。

因此，在链接后，您将获得＆＃34; 未解析的外部符号＆＃34;错误。

<强>解决方案：

将它们全部放入或放出extern C

c / c ++集成和编译 - 即使定义了未解析的外部符号

2 个答案: