[vlc-devel] main libvlc error: input item... when calling libvlc_release

[jboileau]

Hi Pierre,

>(Just a note, you may want to use also the "--ignore-config" vlc
>option, so that local vlc config won't interfer)

Good tip, will use it from now on. Thanks.

I have been doing some more tests this morning. And in response to your inquiry:

> > Is the problem really related to the number of options, or to the
> > options used?

Here is the behaviour this morning:

1. Without any call to libvlc_media_add_option: assert on
libvlc_media_player_stop. (internals->i_refcount > 0 in misc/objects.c
line 845)
2. With only the option (":sout=#duplicate{dst=display}") assert on
libvlc_release. (in file libvlc-common.c line 1008)
3. With only the option (":sout-all") assert on
libvlc_media_player_stop. (internals->i_refcount > 0 in misc/objects.c
line 845)
4. With both options assert on libvlc_release. (in file
libvlc-common.c line 1008)
All above tests but with 2 more options added (test 1 is with these 2
new options only): same results always.

the two new options where: ":sout-ts-es-id-pid" and ":sout-ts-use-key-frames"

I have also tried removing some of the options to libvlc_new keeping
only the path to plugins and "--intf=dummy". I get the same results
except that assert "internals->i_refcount > 0 in misc/objects.c line
845" are now access violations.

Strange, I could get it to work properly on Friday without the
(":sout=#duplicate{dst=display}") but now I get the i_refcount assert
in this case!

I hope this helps. It does not seem to be related to the number of
options but more with the option (":sout=#duplicate{dst=display}").
Let me know of any other test you can think of to help pin point the
problem. Also, if you come to the conclusion that its not a libvlc
problem but a core problem, who would be the correct developer to
contact?

I will download the latest build and give it a spin. See if the
behavior is any different. I will also reboot and try again, I can't
understand why I am getting these errors on stop that I didn't have on
Friday.

Jacques

On Sat, May 10, 2008 at 10:35 AM, jboileau <jboileau at gmail.com> wrote:
> Thanks Pierre,
>
> I will do further tests monday morning. I will try more and different
> options and see if there is some options that are more problematic
> than others. I'll report back my findings.
>
> Thanks for your help,
>
> Jacques
>
>
> On 5/10/08, Pierre d'Herbemont <pdherbemont at free.fr> wrote:
> > Hi Jacques,
> >
> > On May 9, 2008, at 5:56 PM, jboileau wrote:
> >
> > > If I don't add options or if I only add
> > > ":sout-all" I don't have a problem, but if I add both options then it
> > > crashes when I do the libvlc_release.
> >
> > Is the problem really related to the number of options, or to the
> > options used?
> >
> > > int   lNbOptions = 9;
> > > const char* lOptions[] =
> > >       {"C:\\Program Files\\VideoLAN\\VLC-0.9.0\\plugins",
> > >       "--intf=dummy",
> > >       "--no-media-library",
> > >       "--no-one-instance",
> > >       "--reset-plugins-cache",
> > >       "--no-stats",
> > >       "--no-video-title-show",
> > >       "-vvv",
> > >       "--file-logging"};
> >
> > (Just a note, you may want to use also the "--ignore-config" vlc
> > option, so that local vlc config won't interfer)
> >
> > > aLibVLCMedia = libvlc_media_new(aLibVLCInstance,
> > > "c:\\projects\\libvlctest\\debug\\test.avi", &lLibVLCException);
> > > libvlc_media_add_option(aLibVLCMedia, ":sout=#duplicate{dst=display}",
> > > &lLibVLCException);
> >
> > I do believe that this option is the issue. What may be happening is
> > that the input thread is not properly halted.
> >
> > So this could be not directly related to the libvlc-control layer, but
> > probably to vlc's core. It would be nice to investigate more...
> >
> > Pierre.
> > _______________________________________________
> > vlc-devel mailing list
> > To unsubscribe or modify your subscription options:
> > http://mailman.videolan.org/listinfo/vlc-devel
> >
>
>
> --
> Jacques Boileau
>



-- 
Jacques Boileau