Bug 26354 – [Gtk] GStreamer-CRITICAL's (and other warnings) on <video>

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKitGTK

[Adrian Bunk]

- Liferea 1.6
- WekKitGtk+ 1.1.9
- Debian unstable, only installed GStreamer packages are libgstreamer-plugins-base0.10-0 and libgstreamer0.10-0
- viewing http://macslow.net/?p=278 (contains an ogg/theora <video>)

I'll attach the complete log of the warnings and the trace of the first warning.

[Adrian Bunk]

Created attachment 31210 [details]
warnings displayed

[Adrian Bunk]

Created attachment 31211 [details]
backtrace of the first GStreamer-CRITICAL

[Adrian Bunk]

Note:

I do not expect to see the video in this case, but some proper error handling.

[Philippe Normand]

With git WebKitGtk+ and git GStreamer I don't see any warning, either when the page loads or when I play the video.

[Adrian Bunk]

(In reply to comment #4)
> With git WebKitGtk+ and git GStreamer I don't see any warning, either when the
> page loads or when I play the video.

When you are able to play the video you are obviously outside the scope of this bug.

Please remove all GStreamer plugins except for the base ones and retry.

[Philippe Normand]

(In reply to comment #5)
> (In reply to comment #4)
> > With git WebKitGtk+ and git GStreamer I don't see any warning, either when the
> > page loads or when I play the video.
> 
> When you are able to play the video you are obviously outside the scope of this
> bug.
> 
> Please remove all GStreamer plugins except for the base ones and retry.

If I remove the gstreamer ogg plugin the page in MacSlow's blog still loads fine without gst warning. Can you test again please? I think this bug can be closed.

[Adrian Bunk]

(In reply to comment #6)
> (In reply to comment #5)
> > (In reply to comment #4)
> > > With git WebKitGtk+ and git GStreamer I don't see any warning, either when the
> > > page loads or when I play the video.
> > 
> > When you are able to play the video you are obviously outside the scope of this
> > bug.
> > 
> > Please remove all GStreamer plugins except for the base ones and retry.
> 
> If I remove the gstreamer ogg plugin the page in MacSlow's blog still loads
> fine without gst warning. Can you test again please? I think this bug can be
> closed.

No, it is still present with 1.1.16.

You were trying to remove the wrong plugin.

/usr/lib/gstreamer-0.10/libgstplaybin.so (just look at the trace in comment #2) is the file you have to remove for reproducing this bug.

[Philippe Normand]

Created attachment 43757 [details]
Implemented MediaPlayerPrivate::isAvailable by checking the presence of the playbin2 GStreamer element.

[Gustavo Noronha (kov)]

Comment on attachment 43757 [details]
Implemented MediaPlayerPrivate::isAvailable by checking the presence of the playbin2 GStreamer element.

Good. Do you need someone to commit this, or you have commit privileges by now?

Some nits:

 163     bool available = false;
 164     GstElementFactory* factory;
 165 
 166     do_gst_init();
 167     factory = gst_element_factory_find("playbin2");

factory should only be declared here, where it is first used. Perhaps you can also lose the available variable, and just return true inside the if (factory) check, returning false otherwise.

[Philippe Normand]

(In reply to comment #9)
> (From update of attachment 43757 [details])
> Good. Do you need someone to commit this, or you have commit privileges by now?
> 

Received the invitation today, might take some time for the paper work to be done ;)

> Some nits:
> 
>  163     bool available = false;
>  164     GstElementFactory* factory;
>  165 
>  166     do_gst_init();
>  167     factory = gst_element_factory_find("playbin2");
> 
> factory should only be declared here, where it is first used. Perhaps you can
> also lose the available variable, and just return true inside the if (factory)
> check, returning false otherwise.

I used a factory variable because the return value of gst_element_factory_find needs to be unreffed after use.

Do you want me to send a new patch?

[Gustavo Noronha (kov)]

(In reply to comment #10)

> > Some nits:
> > 
> >  163     bool available = false;
> >  164     GstElementFactory* factory;
> >  165 
> >  166     do_gst_init();
> >  167     factory = gst_element_factory_find("playbin2");
> > 
> > factory should only be declared here, where it is first used. Perhaps you can
> > also lose the available variable, and just return true inside the if (factory)
> > check, returning false otherwise.
> 
> I used a factory variable because the return value of gst_element_factory_find
> needs to be unreffed after use.

Yeah. You do need factory. What you could do to it is use GOwnPtr.

The variable I say you should lose is 'bool available'. You can just return true inside the if that checks for factory.

If you can't commit, then yeah, please upload a new patch =)

[Philippe Normand]

Created attachment 43763 [details]
Implemented MediaPlayerPrivate::isAvailable by checking the presence of the playbin2 GStreamer element.

[Philippe Normand]

Created attachment 43764 [details]
Implemented MediaPlayerPrivate::isAvailable by checking the presence of the playbin2 GStreamer element.

[Gustavo Noronha (kov)]

Comment on attachment 43764 [details]
Implemented MediaPlayerPrivate::isAvailable by checking the presence of the playbin2 GStreamer element.

Thanks.

[WebKit Commit Bot]

Comment on attachment 43764 [details]
Implemented MediaPlayerPrivate::isAvailable by checking the presence of the playbin2 GStreamer element.

Clearing flags on attachment: 43764

Committed r51343: <http://trac.webkit.org/changeset/51343>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.