[FFmpeg-devel] Question on libbrokenos support

Måns Rullgård mans
Mon Feb 4 12:23:36 CET 2008

Luca Abeni wrote:
> Hi all,
> first of all, sorry for the long delay
> Michael Niedermayer wrote:
>> On Wed, Jan 09, 2008 at 11:48:32AM +0100, Luca Abeni wrote:
> [...]
>>> Here, I just want to clarify that my proposal about #include_next did not
>>> imply
>>> to install libos headers in standard paths: a user would have needed to
>>> specify
>>> "-I <libos install path>/..." to include such headers, so I think my proposal
>>> did not break the whole system.
>> hmm
>> theres also the option of using
>> gcc -include os_support.h
> I tried this solution, but it unfortunately has problems: in this way,
> os_support.h is included in all files, and it has to include some
> system-dependent headers (for example, winsock2.h and ws2tcpip.h for
> win32 targets) which sometime define symbols and names used by
> os-independent files in libavcodec.
> If I remember well, windows headers end up with a "#define near",
> which break the compilation of some unrelated files (the jpeg encoder,
> I think).
> So, I believe that forcing the inclusion of an "os_support.h" in all the
> files is not a good solution.
> On the other hand, it seems to me that our configure script is already
> using the "-E" gcc's option. So, such option can be used to find the
> location of the system headers, and to copy them in lbos as suggested by
> Michael. (it was objected that this solution is not portable, but if
> ffmpeg's configure script is already using it I think it can be used by
> lbos too, right?).

-E is a POSIX standard compiler flag.

M?ns Rullg?rd
mans at mansr.com

More information about the ffmpeg-devel mailing list