Message ID | alpine.DEB.2.10.1411211853560.2475@digraph.polyomino.org.uk |
---|---|
State | New |
Headers | show |
That change is fine, but I'd be more comfortable if we also changed that fdopen reference rather than implicitly assuming we'll have the internal macro forever. perror is already using _IO_* calls, so it can just use the internal name directly. Thanks, Roland
On Fri, 21 Nov 2014, Roland McGrath wrote: > That change is fine, but I'd be more comfortable if we also changed that > fdopen reference rather than implicitly assuming we'll have the internal > macro forever. perror is already using _IO_* calls, so it can just use If the internal macro is removed (generally or for a particular configuration), failures in the linknamespace tests will make it immediately obvious that this fdopen call needs to change.
diff --git a/conform/Makefile b/conform/Makefile index cf54722..4a498e4 100644 --- a/conform/Makefile +++ b/conform/Makefile @@ -325,12 +325,9 @@ test-xfail-XOPEN2K/ndbm.h/linknamespace = yes test-xfail-XOPEN2K8/ndbm.h/linknamespace = yes # Unsorted expected failures. -test-xfail-ISO/stdio.h/linknamespace = yes test-xfail-ISO/time.h/linknamespace = yes test-xfail-ISO99/ctype.h/linknamespace = yes -test-xfail-ISO99/stdio.h/linknamespace = yes test-xfail-ISO11/ctype.h/linknamespace = yes -test-xfail-ISO11/stdio.h/linknamespace = yes test-xfail-XPG3/ctype.h/linknamespace = yes test-xfail-XPG3/fnmatch.h/linknamespace = yes test-xfail-XPG3/glob.h/linknamespace = yes diff --git a/stdio-common/perror.c b/stdio-common/perror.c index 4b47015..7de83e4 100644 --- a/stdio-common/perror.c +++ b/stdio-common/perror.c @@ -56,7 +56,7 @@ perror (const char *s) oriented yet? In this case we'll create a new stream which is using the same underlying file descriptor. */ if (__builtin_expect (_IO_fwide (stderr, 0) != 0, 1) - || (fd = fileno (stderr)) == -1 + || (fd = __fileno (stderr)) == -1 || (fd = __dup (fd)) == -1 || (fp = fdopen (fd, "w+")) == NULL) {