Bruno Haible wrote:
>> FAIL: test-fgetc.exe
>> FAIL: test-fputc.exe
>> FAIL: test-fread.exe
>> FAIL: test-fwrite.exe
>>
>> All these fail on a stream whose file descriptor has been closed. Most
>> libraries won't do such a thing. Therefore I don't find it worth to fix
>> this. Just document it.
>
> But two of the tests still fail, and the other two succeed only accidentally.
> This patch makes the 4 tests pass:
Hi Bruno,
I noticed that the latest from gnulib's tests gets some new warnings
when compiled on Fedora 15:
test-fread.c: In function 'main':
test-fread.c:43:3: error: implicit declaration of function
'gl_msvc_inval_ensure_handler' [-Werror=implicit-function-declaration]
cc1: all warnings being treated as errors
test-fputc.c: In function 'main':
test-fputc.c:43:3: error: implicit declaration of function
'gl_msvc_inval_ensure_handler' [-Werror=implicit-function-declaration]
cc1: all warnings being treated as errors
test-fwrite.c: In function 'main':
test-fwrite.c:43:3: error: implicit declaration of function
'gl_msvc_inval_ensure_handler' [-Werror=implicit-function-declaration]
cc1: all warnings being treated as errors
make[3]: *** [test-fputc.o] Error 1
make[3]: *** Waiting for unfinished jobs....
make[3]: *** [test-fread.o] Error 1
make[3]: *** [test-fwrite.o] Error 1
CC test-strings.o
test-fgetc.c: In function 'main':
test-fgetc.c:43:3: error: implicit declaration of function
'gl_msvc_inval_ensure_handler' [-Werror=implicit-function-declaration]
cc1: all warnings being treated as errors
That's because the guard in each test-*.c file is weaker
than the guards in msvc-inval.h. Thus, the use is exposed
while the declaration is not.
>From msvc-inval.h:
#define DEFAULT_HANDLING 0
#define HAIRY_LIBRARY_HANDLING 1
#define SANE_LIBRARY_HANDLING 2
#if HAVE_MSVC_INVALID_PARAMETER_HANDLER \
&& !(MSVC_INVALID_PARAMETER_HANDLING == SANE_LIBRARY_HANDLING)
/* A native Windows platform with the "invalid parameter handler" concept,
and either DEFAULT_HANDLING or HAIRY_LIBRARY_HANDLING. */
# if MSVC_INVALID_PARAMETER_HANDLING == DEFAULT_HANDLING
/* Default handling. */
# ifdef __cplusplus
extern "C" {
# endif
/* Ensure that the invalid parameter handler in installed that just returns.
Because we assume no other part of the program installs a different
invalid parameter handler, this solution is multithread-safe. */
extern void gl_msvc_inval_ensure_handler (void);
# ifdef __cplusplus
}
# endif
>From test-fread.c:
#if MSVC_INVALID_PARAMETER_HANDLING == DEFAULT_HANDLING
gl_msvc_inval_ensure_handler ();
#endif
To kludge around it, I built with this:
make check CFLAGS=-DMSVC_INVALID_PARAMETER_HANDLING=99
Wouldn't it be better to make the commonly-used-in-application-code
gl_msvc_inval_ensure_handler ();
a no-op when possible, so that applications don't need to guard
uses of it with #if directives like those above?
