[MPlayer-dev-eng] DeckLink driver

[Reimar Döffinger]

On Mon, Sep 19, 2011 at 11:11:02PM +0200, Diego Biurrun wrote:
> On Mon, Sep 19, 2011 at 01:20:43PM -0600, Deron Kazmaier wrote:
> > On 9/17/11 3:08 PM, Diego Biurrun wrote:
> > >On Sat, Sep 17, 2011 at 11:56:47AM -0600, Deron Kazmaier wrote:
> > >>On 9/17/11 10:58 AM, Reimar Döffinger wrote:
> > >>>On Sat, Sep 17, 2011 at 10:49:04AM -0600, Deron Kazmaier wrote:
> > >>>>I've updated the code to compile (well, one niggly that I don't
> > >>>>fully understand but it is with a Delete which hopefully will at
> > >>>>worst just leave memory unfreed at the end) but I couldn't get it
> > >>>>too link. The error messages certainly made me believe that the
> > >>>>problem is simply that the C++ libraries are not linked in. Just
> > >>>>using g++ instead of gcc seems to fix it (I copy/pasted the link
> > >>>>line, replacing gcc with g++), but what would be the proper way to
> > >>>>handle this in the config? Obviously the average user would not
> > >>>>enable/compile the decklink drivers and so it would be crazy to use
> > >>>>g++ all the time to link it.
> > >>>Well, you are really very vague and don't give us any relevant
> > >>>information at all.
> > >>>But usually that would be because you're missing -lstdc++
> > >>Sorry, I may have given too much info.
> > >No, you have given too little.  Please send the patch and show us
> > >the error message.
> > 
> > Sorry. I was talking about the background. I never know how much to
> > write before eyes start glazing over and people reach for the delete
> > key :-)
> > 
> > I have this compiling and linking and running by hand (I simply
> > copy/paste the final mplayer link line, replacing cc with g++), but
> > it would be nice if configure could take care of it. The actual
> > DeckLink driver patch was posted on this list and can be seen here:
> > 
> > http://mplayerhq.hu/pipermail/mplayer-dev-eng/2009-August/062097.html
> > 
> > patch url http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/attachments/20090817/0a55034d/attachment.diff.
> 
> Why not port this to C and be done with it?  I honestly don't think we
> should accept C++ ao/vo drivers.

Because accessing a C++ only library from C is madness of the kind where
you don't want to go?
Now I'd be all for beating decklink with a cluestick for making it
impossible to use their devices from pure C, particularly since I
don't really see them make any good use of C++, but I don't see
that having a large chance of success.