[FFmpeg-devel] Need information about stripping FFMpeg Library

[sandeep virdi]

On Tue, Dec 22, 2009 at 6:13 PM, Diego Biurrun <diego at biurrun.de> wrote:

>  On Tue, Dec 22, 2009 at 06:02:21PM +0530, sandeep virdi wrote:
> > On Tue, Dec 22, 2009 at 4:42 PM, Diego Biurrun <diego at biurrun.de> wrote:
> >
> > > On Tue, Dec 22, 2009 at 04:28:16PM +0530, sandeep virdi wrote:
> > > >
> > > > Any information will be highly appreciated.
> > >
> > > http://www.ffmpeg.org/developer.html
> > >
> > I went through the link pointed by you, but in my case as I previously
> > mentioned that I had removed almost everything except what is needed to
> > decode H264 files. That means I have changed the functionality of FFMpeg
> > lib. So I was wondering as to how could I submit this under normal FFMpeg
> > distribution. I am not really aware of how open source projects are
> handled
> > since this is first time for me. So since my changes are so huge as I
> have
> > removed most of the code files all together, I was wondering as to if I
> can
> > publish my changes as a new open source project, something like
> > "Lean-FFMpeg" and if so then what will be the procedure. As the code is a
> > subset of FFMpeg, I think that this is the ideal place to put this lean
> > FFMpeg code.
>
> FFmpeg already supports removing almost all functionality except certain
> specific parts.  How does what you did differ from passing
>
>  --disable-decoders --enable-decoder=h264
>
> (and probably a few options more) to configure?
>
> Starting a new project sounds like a bad idea.  You claim to have little
> experience with that and it would duplicate a lot of work.
>
> Diego
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at mplayerhq.hu
> https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel
>


Sorry for top-posting in my previous mail. :)
Thanks for the information about using the configure script. Yeah that seems
to be a better idea so as to only have code for H264 decoder in final
binary. To give a brief info about why I thought of stripping FFmpeg was,
that our security team found that there were few known vulnerabilities in
FFMpeg and hence they were wary of shipping FFmpeg(in form of  Dlls) along
with our code. So that's why I thought of removing unwanted code for my
case.

But there is still the issue of reading of encoded stream. I had seen in the
code that with FFMpeg we can have several protocols to input. In my case, I
will be having one .264 encoded frame in memory and only after this encoded
frame is decoded, that the next encoded frame can come. So its like at any
given time the decoder will have only one frame in memory to decode. I
couldn't find a protocol which can handle. In-fact there is no protocol in
FFmpeg which allows for reading input data from memory. Can you point
me/give hint out as to how this can be achieved using existing FFMpeg
methods.

Thanks for your time and Merry Christmas :)
Regards
Sandeep