Commit Message
sys/cdefs.h has a macro __long_double_t used in two places in glibc.
long double is a standard part of C since C89; there is no need for
such an alias for it. This patch removes that macro and uses long
double directly everywhere. As an implementation-namespace,
undocumented symbol, it should not be considered part of the API for
users, and codesearch.debian.net shows no sign of it being used
outside glibc in a way that would break with this patch.
Tested for x86_64.
2017-08-04 Joseph Myers <joseph@codesourcery.com>
* misc/sys/cdefs.h (__long_double_t): Remove.
* stdio-common/printf_fp.c (__printf_fp_l): Use long double
instead of __long_double_t,
* stdlib/strfmon_l.c (__vstrfmon_l): Likewise.
Comments
On Mon, Aug 7, 2017 at 3:53 PM, Joseph Myers <joseph@codesourcery.com> wrote:
> Now committed.
Sorry, I meant to respond to this.
I would bet that __long_double_t was inherited from the pre-C89 BSD
<sys/cdefs.h>, and I don't believe we have ever encouraged direct use
of that header, so this seems fine to me.
zw
@@ -102,7 +102,6 @@
/* This is not a typedef so `const __ptr_t' does the right thing. */
#define __ptr_t void *
-#define __long_double_t long double
/* C++ needs to know that types and declarations are C, not C++. */
@@ -217,7 +217,7 @@ __printf_fp_l (FILE *fp, locale_t loc,
union
{
double dbl;
- __long_double_t ldbl;
+ long double ldbl;
#if __HAVE_DISTINCT_FLOAT128
_Float128 f128;
#endif
@@ -100,7 +100,7 @@ __vstrfmon_l (char *s, size_t maxsize, locale_t loc, const char *format,
union
{
double dbl;
- __long_double_t ldbl;
+ long double ldbl;
}
fpnum;
int int_format;