Bug 16985 – Webkit video element does not repaint video with Gtklauncher

• Status: RESOLVED
• Resolution: DUPLICATE
• Component: WebKitGTK

[Rune B]

With latest version of webkit svn checkout I made a test where I built Webkit with ./configure --enable-video --enable.svg --enable-debug

I then made a test html page with the video element included.

Webkit only shows the first frame, but audio continue as normal.

However, if I resize, maximize, minimize etc I trigger a repaint of the video and it jumps to the current frame. Then it only shows that frame until I e.g resize again.

I am using gstreamer as a backend on a Debian Testing box.

I tested this both with a .mov and with an .ogg

html + javascript code I am using

<script>
function playPause() {
  var myVideo = document.getElementsByTagName('video')[0];
  if (myVideo.paused)
    myVideo.play();
  else
    myVideo.pause();
}
</script>

<p><video src=files/movie.mov autoplay></video> <br/><input type=button onclick="playPause()" value="Play/Pause"></p>

[Rune B]

Created attachment 18621 [details]
Screenshot

Screenshot of Gtklauncher

[Alp Toker]

We need to track this change that was made to WebCore and the Mac platform code:

  http://trac.webkit.org/projects/webkit/changeset/28610

For GStreamer, this probably means adding a call to repaint() in WebCore/platform/graphics/gtk/VideoSinkGStreamer.cpp

[Sébastien Roret]

I tested a timer in MediaPlayerPrivate that fire repaint() every 1/25th of a second, and the video is smooth.
Unfortunately, I have no idea how to ask gstreamer to be notified for each new frame, in order to implement it properly...

[Zan Dobersek]

Created attachment 27292 [details]
Repaint media player when video sink allows it

A new signal is introduced in WebKitVideoSink, which is emitted when sink's surface is updated and the repaint is possible. Callback then calls WebCore::MediaPlayerPrivate::repaint and the repaint is done properly.

All in all, it fixes this bug's problem.

[Gustavo Noronha (kov)]

I think I have duplicated some of this work =(. I have landed a very similar patch earlier this year (bug #24638). Sorry Zan =(.

*** This bug has been marked as a duplicate of 24638 ***