[PATCH 2/2] i18n: avoid conflict with ngettext from libintl
- From
Jonathan Nieder <jrnieder@gmail.com>
- Date
- Mar 9, 2011, 10:52 UTC
- Message-ID
- <20110309105236.GC30980@elie>
- In-Reply-To
- <20110309103104.GA30980@elie>
Although git itself would not be using "#include <libintl.h>" anywhere in a NO_GETTEXT build, git's gettext.h is meant to tolerate prior declarations from libintl, to prepare for a scenario in which some system or compat/ header decides to start including libintl. GNU libintl.h defines ngettext as a macro when __OPTIMIZE__ is defined, so take care to "#undef ngettext" if it was defined for us.
To avoid having to worry about a conflicting ngettext symbol when libintl is part of libc, also rename the no-op ngettext stub to git_ngettext and make ngettext a macro referring to it. This is probably never necessary (because git's ngettext is declared "static inline") but it buys peace of mind.
This change does not protect against conflicts due to a header included _after_ git's i18n support (e.g., pthread.h) being the first to pull in libintl. We can deal with that separately if it happens.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com> --- gettext.h | 7 ++++++- 1 files changed, 6 insertions(+), 1 deletions(-)
diff --git a/gettext.h b/gettext.h index 03fb340..a473af4 100644 --- a/gettext.h +++ b/gettext.h @@ -13,6 +13,10 @@ #error "namespace conflict: '_' is pre-defined?" #endif +#ifdef ngettext +#undef ngettext +#endif + #define FORMAT_PRESERVING(n) __attribute__((format_arg(n))) #ifdef GETTEXT_POISON @@ -26,8 +30,9 @@ static inline FORMAT_PRESERVING(1) const char *_(const char *msgid) return use_gettext_poison() ? " GETTEXT POISON " : msgid; } +#define ngettext git_ngettext static inline FORMAT_PRESERVING(1) FORMAT_PRESERVING(2) -const char *ngettext(const char *msgid, const char *plu, unsigned long n) +const char *git_ngettext(const char *msgid, const char *plu, unsigned long n) { if (use_gettext_poison()) return " GETTEXT POISON ";
-- 1.7.4.1