[FFmpeg-devel] [PATCH] Fix inclusion in pixdesc.h of the non-public header intreadwrite.h

[Michael Niedermayer]

On Tue, Feb 16, 2010 at 12:28:27AM +0000, M?ns Rullg?rd wrote:
> Michael Niedermayer <michaelni at gmx.at> writes:
> 
> > On Mon, Feb 15, 2010 at 10:47:15PM +0000, M?ns Rullg?rd wrote:
> >> Michael Niedermayer <michaelni at gmx.at> writes:
> >> 
> >> > On Mon, Feb 15, 2010 at 07:24:57PM +0000, M?ns Rullg?rd wrote:
> >> >> Michael Niedermayer <michaelni at gmx.at> writes:
> >> >> 
> >> >> > On Mon, Feb 15, 2010 at 06:37:19PM +0000, M?ns Rullg?rd wrote:
> >> >> >> Michael Niedermayer <michaelni at gmx.at> writes:
> >> >> >> 
> >> >> >> > On Mon, Feb 15, 2010 at 01:00:42PM +0000, M?ns Rullg?rd wrote:
> >> >> >> >> Michael Niedermayer <michaelni at gmx.at> writes:
> >> >> >> >> 
> >> >> >> >> > On Sun, Feb 14, 2010 at 11:39:35PM +0100, Stefano Sabatini wrote:
[...]
> > The code is used (mplayer is one example)
> > The code works very well (mplayer is the proof)
> 
> MPlayer sets up all the necessary definitions for it to work.  Others
> are of course free to include their own ffmpeg copy just like MPlayer
> does.

The thing i disagree with you about is
why does that need a copy?
put the whole header under
#ifndef I_did_RTFM_and_i_did_set_up_all_platform_and_compiler_specific_defines
to me physical copying (or a symlink) simply seems a very annoying solution


> 
> > We can make it work even better by checking the used compiler and cpu
> > for the cases that the user compiles lib & app with different compilers
> > or exchanged his cpu in the meantime.
> 
> No, we cannot.  

Which complier that we currently have specific code in intreadwrite.h is
undetectable though a simple #if(def) ?
gcc, and the compatible compilers like icc can be detected ...


> Are you suggesting we install the configure script and
> rerun it every time the header is used?  (Not that _that_ is possible
> either.)
> 

> > Nothing you can do will prevent mplayer and others from using this code
> > I have in the past said that the header should be properly installed,
> > its a matter of time until distros will apply patches for this as more
> > packages start depending on it. (mplayer needs it already)
> 
> Its fault, not mine.

doesnt make a difference to any user of mplayer or any user of libavutil


> 
> > Our choice is just between trying to make it work for most (that are
> > 99.9999%) of the cases or trying to be one of the FOSS philosophers who
> > put their ideas of what is right and wrong above actual useability and
> > their users.
> 
> You're the one putting your ideas of what you'd like above actual reality.

If my ideas are impossible i can accept that but you arent convincing at all
where is the compiler that configure can detect that #if(defs) cant?

Also, applications need the functionality of intreadwrite.h
The only option you suggested is they should fork intreadwrite.h, i agree
this is better than everyone reimplementing it from scratch but it still is
not terribly convenient
Making intreadwrite.h easier to use (copy or not) is surely usefull to some
applications. In this sense i think making it less dependant on configure and
instead use __GNUC__ and similar would make it much easier to copy and use


[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Old school: Use the lowest level language in which you can solve the problem
            conveniently.
New school: Use the highest level language in which the latest supercomputer
            can solve the problem without the user falling asleep waiting.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20100216/3760d847/attachment.pgp>