[FFmpeg-devel] [PATCH v5 2/2] avformat/os_support: Support long file names on Windows

[Martin Storsjö]

On Tue, 24 May 2022, Soft Works wrote:

>> -----Original Message-----
>> From: Martin Storsjö <martin at martin.st>
>> Sent: Tuesday, May 24, 2022 11:23 AM
>> To: FFmpeg development discussions and patches <ffmpeg-devel at ffmpeg.org>
>> Cc: softworkz <softworkz at hotmail.com>; Hendrik Leppkes
>> <h.leppkes at gmail.com>
>> Subject: Re: [FFmpeg-devel] [PATCH v5 2/2] avformat/os_support: Support
>> long file names on Windows
>>
>> On Tue, 24 May 2022, softworkz wrote:
>>
>>> From: softworkz <softworkz at hotmail.com>
>>>
>>> Signed-off-by: softworkz <softworkz at hotmail.com>
>>> ---
>>> libavformat/os_support.h | 16 +++++++++++-----
>>> 1 file changed, 11 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/libavformat/os_support.h b/libavformat/os_support.h
>>> index 5e6b32d2dc..d4c07803a5 100644
>>> --- a/libavformat/os_support.h
>>> +++ b/libavformat/os_support.h
>>> @@ -49,7 +49,13 @@
>>> #  ifdef stat
>>> #   undef stat
>>> #  endif
>>> -#  define stat _stati64
>>> +#  define stat win32_stat
>>> +
>>> +    struct win32_stat
>>> +    {
>>> +        struct _stati64;
>>> +    };
>>
>> Is it possible to work around this issue by doing "#define stat(a,b)"
>> which only should apply on the function, not to the struct?
>
> How could this be possible? A define is only doing string replacements,
> so I wouldn't know how it could be restricted to the function, but
> not the struct.

If unsure about a tool feature, please try it out for yourself. Yes, a 
define is only a string replacement, but a define with parameters only 
matches the string occurs with parenthesis afterwards. Consider this 
example:

$ cat test.c
#define stat(a, b) win32_stat(a, b)

struct stat {
 	int a, b, c;
};

void stat (struct stat *st, const char* filename);

void func(const char* filename) {
 	struct stat st;
 	stat (&st, filename);
}
$ cc -E test.c
# 1 "test.c"
# 1 "<built-in>"
# 1 "<command-line>"
# 31 "<command-line>"
# 1 "/usr/include/stdc-predef.h" 1 3 4
# 32 "<command-line>" 2
# 1 "test.c"


struct stat {
  int a, b, c;
};

void win32_stat(struct stat *st, const char* filename);

void func(const char* filename) {
  struct stat st;
  win32_stat(&st, filename);
}


So here, the stat -> win32_stat rewrite only applied on the function 
declaration and call, but not on the structs.

Not saying that this necessarily is the way forward, but I was just 
mentioning it as a potential option to consider.

>> A safe way forward could be to switch code to just using "struct
>> ff_stat_struct", and define ff_stat_struct to the name of the struct we
>> exepct to use. It's not pretty, and it affects users which no longer can
>> use the default POSIX stat form of the calls
>
> That's why I took the effort to make this work.
>
>> but it would fix the issue
>> of redirecting the struct and function separately, without needing to know
>> what exactly is in the struct (because we really shouldn't be
>> hardcoding/assuming that).
>
> That doesn't apply because the current code already does this:
>
> DEF_FS_FUNCTION2(stat, _wstati64, _stati64, struct stat*)
>
> Which means that it specifically chooses _stati64 which is a public
> MS API:
>
> https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/stat-functions?view=msvc-170
>
> And we know that it takes _stati64 as the parameter.

Yes, there is no disagreement about that part.

> This code:
>
>    struct win32_stat
>    {
>        struct _stati64;
>    };
>
> makes use of a MS compiler feature called "anonymous structures":
>
> https://docs.microsoft.com/en-us/previous-versions/visualstudio/visual-studio-2013/z2cx9y4f(v=vs.120)?redirectedfrom=MSDN
>
> This way, we automatically "inherit" the members of the struct
> (whatever their members would be).

This, as the article itself clearly declares, is a C language extension. 
GCC allows it in mingw mode, but Clang doesn't. (It's possible to use it 
in Clang too if you enable it with -fms-extensions though.)

$ cat test2.c
struct orig_struct {
 	int a, b, c;
};
struct my_struct {
 	struct orig_struct;
};
void set(struct my_struct *st) {
 	st->a = 42;
}
$ clang -target x86_64-w64-mingw32 -c test2.c
test2.c:5:2: warning: declaration does not declare anything [-Wmissing-declarations]
         struct orig_struct;
         ^
test2.c:8:6: error: no member named 'a' in 'struct my_struct'
         st->a = 42;
         ~~  ^
1 warning and 1 error generated.
$ clang -target x86_64-w64-mingw32 -c test2.c -fms-extensions
test2.c:5:2: warning: anonymous structs are a Microsoft extension [-Wmicrosoft-anon-tag]
         struct orig_struct;
         ^
1 warning generated.


// Martin