wcrtomb() broken

[Silvan Scherrer]

this testcase shows the broken behaviour

#include <stdlib.h> #include <stdio.h>

int main() {

char buf[256] = {0}; int rc = wctomb(buf, 0xE4); printf("rc %d char %02x (%c)\n", rc, (unsigned char)buf[0], buf[0]); return 0;

}

[Silvan Scherrer]

i get the following output: rc 1 char e4 (õ)

lang is de_CH

and i expected to get a ä

[Silvan Scherrer]

i hate trac :) code snipped now better formated.

#include <stdlib.h>
#include <stdio.h>

int main()
{
	char buf[256] = {0};
	int rc = wctomb(buf, 0xE4);
	printf("rc %d char %02x (%c)\n", rc, (unsigned char)buf[0], buf[0]);
	return 0;
}

[bird]

I'm a bit surprise you get anything since you're not calling setlocale(LC_ALL, "")... Could you do that and also supply the hex value of the LATIN SMALL LETTER A WITH DIAERESIS character you're expecting.

[Silvan Scherrer]

ok did the test. even with LC_ALL i get the same output. the hex value of LATIN SMALL LETTER A WITH DIAERESIS (ä) is 84

[bird]

What happens is that when setting LC_ALL, LC_CTYPE or LANG to 'de_CH' you get a 'ISO8859-1' codepage, and in that code page LATIN SMALL LETTER A WITH DIAERESIS has the same value as the unicode code point. See ​http://en.wikipedia.org/wiki/ISO/IEC_8859-1 for details.

The codepage you're expecting to use is probably 850 (​http://en.wikipedia.org/wiki/Code_page_850). To get output in that codepage you need to set LC_ALL, LC_CTYPE or LANG to the value 'de_CH.IBM850'.

I've added a testcase in r3756 which shows very clearly what setlocale() does and that the mbtowc <-> wctomb round-trip works correctly. There is no bug here, I think.

[bird]

r3788 changes the setlocale behavior when no codeset is specified, it now defaults to the process code page.