On 12 September 2012 19:57, Paul Eggert <egg...@cs.ucla.edu> wrote:
> On 09/12/2012 11:29 AM, Ed Maste wrote:
>> 2605 locale_t thread_locale = uselocale (NULL);
>
> I'm afraid you may need to be the guy to debug this.
> But if this is returning a value other than LC_GLOBAL_LOCALE
> that is a bug, no? POSIX says uselocale (0) should
> return LC_GLOBAL_LOCALE if no thread-local locale has
> been installed.
This test is setting a thread-local locale:
#if HAVE_NEWLOCALE
/* Check that gl_locale_name considers the thread locale. */
{
locale_t locale = newlocale (LC_ALL_MASK, "fr_FR.UTF-8", NULL);
if (locale != NULL)
{
uselocale (locale);
name = gl_locale_name (LC_CTYPE, "LC_CTYPE");
ASSERT (strcmp (name, "fr_FR.UTF-8") == 0);
I see what's going on here though - now that the xlocale API exists
(as of FreeBSD 9.1) HAVE_NEWLOCALE is true, but there is no FreeBSD
implementation in gl_locale_name_thread_unsafe so the function ends up
as:
gl_locale_name_thread_unsafe (int category, const char *categoryname)
{
{
locale_t thread_locale = uselocale (NULL);
if (thread_locale != LC_GLOBAL_LOCALE)
{
}
}
return NULL;
}
which of course isn't going to work well.
-Ed
