swprintf fails with \xffff unicode characters

Question

The following lines of code running with Visual Studio 2013:

wchar_t test[] = L"\xffff";
wchar_t buf[100];    
int ret = swprintf(&buf[0], 100, L"%ls", &test[0]);

It compiles successfully, but swprintf fails to output this character (res is -1). It's ok with any other character except \xffff. This character can be in any position of the string, and swprintf still fails. What's the problem?

Whenever you're asking about an error message, you need to include the exact text of the error message in your question. — Keith Thompson, Aug 07 '14 at 17:54
It's not a valid Unicode string if it contains the code point U+FFFF. See [Can a valid Unicode string contain FFFF](http://stackoverflow.com/questions/3482683/can-a-valid-unicode-string-contain-ffff-is-java-characteriterator-broken). — Jongware, Aug 07 '14 at 18:09
Is the output being save to a file such that you hex-dump it and know that the function failed/succeeded to output? Or are you looking at a screen with textual output - in this case how do you know `'\xffff'` must print something that is visible? — chux - Reinstate Monica, Aug 07 '14 at 18:21

score 3 · Accepted Answer · answered Aug 07 '14 at 18:42

MSVC/MSVCRT has 16-bit wchar_t and thus uses 0xffff as the value of WEOF, thereby precluding it from being a valid character and necessitating that it cause an error when used like this. This does not really preclude use of a Unicode character, since U+FFFF is a non-character in Unicode, but it's still rather undesirable behavior since UTFs are supposed to handle all Unicode scalar values, not just characters.

The only "fix" is to use a proper implementation with 32-bit wchar_t (I think cygwin should work on windows but I'm not sure; otherwise, any unix-like system).

swprintf fails with \xffff unicode characters

1 Answers1