[Ffmpeg-devel] [PATCH] MingW RTSP support

[angustia at arrozcru.no-ip.org]

Hello,

Quoting Diego Biurrun <diego at biurrun.de>:

> On Mon, Oct 30, 2006 at 08:50:11PM +0100, Diego Biurrun wrote:
>> On Mon, Apr 03, 2006 at 01:50:45PM -0500, Michael A. Kohn wrote:
>> >
>> > > On Sun, Apr 02, 2006 at 09:58:10AM -0500, Michael A. Kohn wrote:
>> > >
>> > > Yes. Somehow BeOS lusers get away with this stuff because their OS
>> > > isn't considered as "lame" as windows by most developers, but I don't
>> > > distinguish. One broken, standards-violating, proprietary OS is as bad
>> > > as another..
>> >
>> > Okay, I'm going to try one more time I think.  This patch is just
>> > like the others, except I changed the way the headers look.  I downloaded
>> > a copy of BeOS just to make sure these changes don't change the way it
>> > compiles.  The headers that used to look like this:
>> >
>> > #include "avformat.h"
>> > #include <unistd.h>
>> > #include <sys/types.h>
>> > #include <sys/socket.h>
>> > #include <netinet/in.h>
>> > #if defined(__BEOS__) || defined(__INNOTEK_LIBC__)
>> > typedef int socklen_t;
>> > #endif
>> > #ifndef __BEOS__
>> > # include <arpa/inet.h>
>> > #else
>> > # include "barpainet.h"
>> > #endif
>> > #include <netdb.h>
>> > #include <sys/time.h>
>> > #include <fcntl.h>
>> >
>> > now look like this (and mingw compiles and works):
>> >
>> > #include "avformat.h"
>> > #include <unistd.h>
>> > #include <sys/types.h>
>> > #include <sys/time.h>
>> > #include <fcntl.h>
>> > #if !defined(__MINGW32__) && !defined(__BEOS__)
>> > #include <sys/socket.h>
>> > #include <netinet/in.h>
>> > #include <netdb.h>
>> > #include <arpa/inet.h>
>> > #endif
>> >
>> > I think this is a nice comprimise.. the code looks cleaner now and
>> > both broken OS's compile...
>> >
>> > the only problem with BeOS (which I guess doesn't matter anyway) is
>> > SO_ERROR doesn't exist so without commenting or #defining out, it won't
>> > compile anyway... maybe no one even uses ffmpeg under BeOS so no one
>> > noticed?
>> >
>> > One other extra thing I did... I added this these lines for UINT32_MAX
>> > and INT32_MAX in os_support.h:
>> >
>> > #ifndef UINT32_MAX
>> > #define UINT32_MAX (4294967295U)
>> > #endif
>> >
>> > BeOS required this.. and I believe OpenBSD requires it too.  So these
>> > lines might be a good thing whether this patch is accepted or not..
>>
>> I've looked at this patch again, here is a slightly cleaned-up version.
>> Is this OK to commit?
>
> /* no comment */
>

I couldn't get it to apply properly. Seems like the last hunk of tcp.c  
was diff'd from withing ffmpeg, and the previous part of the patch was  
from ffmpeg's parent folder.

How about changing "Win32isStupid" to "ret" and "microsuck_GUID" to  
"microsoft_GUID"? Seriously, it looks like an 8th grader thinking he's  
so damn cool because he wrote "shit" on his homework...

Ramiro Polla