mbstowcs

[John E]

For some strange reason I always assumed that mbstowcs (and its associated functions) were Microsoft specific but when I googled just now, it looked like those functions are available for other platforms too.

So what kind of wide characters would we get on Linux or OS-X? Would it give us UTF-8 on those platforms or would we still get Windows style (2-byte) characters?

[Codeplug]

The size and encoding of a wchar_t is implementation defined. On Windows OS's, a wchar_t is always 2 bytes and uses UTF16-LE. On many *nix flavors, wchar_t is 4 bytes and uses UTF32 with native endianess.

The narrow encoding that mbstowcs assumes comes form the LC_CTYPE of the current locale.

gg

[John E]

Earlier today I was looking into LC_CTYPE but although I could understand how it might affect character conversions (toupper() / tolower() etc) I couldn't quite understand what affect it has on character representations.

For example on Windows, 'non-wide' characters are usually represented as single bytes. The actual character printed depends on the user's code page. On Linux, 'non-wide' characters are usually UTF-8. So what does mbstowcs() do on a Linux system? Does it convert variable width UTF-8 characters into fixed with (32-bit) wide characters?

Interesting stuf...!

[Igor Vartanov]

For example on Windows, 'non-wide' characters are usually represented as single bytes

Usually yes. But the same way in Windows non-wide character is called MBC which is Multi Byte Character.

[John E]

Thanks Igor. So does that effectively answer my question...

what does mbstowcs() do on a Linux system? Does it convert variable width UTF-8 characters into fixed with (32-bit) wide characters?

Given that UTF-8 is the most common (multi byte) character representation now on Linux, I guess the answer is "yes" ?

[Igor Vartanov]

I'd say "yes" too, but believe that the most effective way to find that is to build a demo app an inspect the memory. Much more reliable compared to asking about Linux on Windows forum.