On 3 May 2017 at 17:26, Liu Hao wrote:
> On 2017/5/4 0:04, Papa wrote:
>>
>> Yes, absolutely right. That's what I thought as well.
>> Now, who should look at this problem, MinGW64 or Gnu, who should I
>> contact to report this bug?
>
> Please make sure it is really a bug before submitting it, by printing the
> first byte of the string in question in your second example, in order to
> make sure it is really a null-terminator (`--enable-fully-dynamic-string`
> may cause trouble here, see PR16612):
>
> ```c++
> void jme::Locale::setGlobalLocale(const std::string& str){
> std::string tmp;
> __builtin_printf("str[0] = %d\n", str.data()[0]);
> tmp = std::setlocale(LC_ALL, str.data() );
> }
> ```
>
> I ask this mostly because the function `setlocale()` is provided by MSVCRT
> and is totally opaque to MinGW-w64. And I can't reproduce the problem:
Good to know. That's why I qualified my guess with ***Assuming the
content of str is "",***. Because otherwise it's just another case of
passing an unsupported argument to setlocale.
