[Ffmpeg-devel] Call for volunteers to write NUT de/muxer for libavformat

[The Wanderer]

Oded Shimon wrote:

> On Tue, Oct 31, 2006 at 08:25:25PM -0500, The Wanderer wrote:
> 
>> Oded Shimon wrote:
>> 
>>> On Mon, Oct 30, 2006 at 10:39:49AM -0800, Mike Melanson wrote:
>> 
>>>> However, are their any sample files I can look at to get me
>>>> started?
>>> 
>>> Make them yourself from either ogg vorbis files or mpeg-4+mp3+avi
>>>  files using nutmerge - svn co svn://svn.mplayerhq.hu/nut/
>> 
>> Is this really appropriate given that nutmerge itself, as obtained
>> via that command line, prints a big fat warning to *not* use it for
>> anything other than testing purposes?
> 
> This warning will be removed once:
> 1. nutmerge rejects all files it can't handle (this is almost the
> current situation)
> 2. libnut "version" will be pushed...

Acknowledged. Just wanted to make sure about the situation.

>> On a somewhat unrelated note, I have an AVI with XviD (which is,
>> technically, MPEG-4) video and MPEG-1 layer 3 audio on which
>> nutmerge exits with code 05 after printing the warning message,
>> without writing even one byte to the output file. After spending
>> perhaps fifteen minutes learning things I'm surprised I didn't know
>> about gdb, I still haven't been able to track down the place in the
>> code where that value is set, but could make the failing file
>> available if it were desired.
> 
> Can I have this file? I looked up error 5, that's a really weird
> error to get...

I apologize - I do not in fact have such a file. When I first wrote that
paragraph I'd forgotten about the MP3-audio requirement (the error-5
file has uncompressed PCM audio), and when I found a second file which
does fit those specifications and which exited with only the warning
banner (at which point I added the "and MPEG-1 layer 3 audio") I somehow
didn't remember to test its return value explicitly; it turns out to
return code 01.

If you also want that second file, I can provide it.

> BTW, 'grep -r "return 5" .'

Gah. I didn't bother with anything like that for the simple reason that
I'm much more accustomed to returning variables, or at least macro
names, than returning explicit values - combined with the fact that
variables can be altered arithmetically rather than simply assigned to.
I was attempting to track the variable returned by main, "err", back
through the code until I found which test was failing which would cause
the program to begin the cascade towards exiting - and *then* track back
further until I found where that value was set, and then figure out why.

-- 
       The Wanderer

Warning: Simply because I argue an issue does not mean I agree with any
side of it.

Secrecy is the beginning of tyranny.