[Webkit-unassigned] [Bug 35992] Crash at MediaPlayer::duration()

Thu Mar 11 11:50:20 PST 2010

https://bugs.webkit.org/show_bug.cgi?id=35992

--- Comment #4 from Hin-Chung Lam <hclam at google.com>  2010-03-11 11:50:20 PST ---
(In reply to comment #3)
> (From update of attachment 50463 [details])
> >  float HTMLMediaElement::duration() const
> >  {
> > -    if (m_readyState >= HAVE_METADATA)
> > +    if (m_player && m_readyState >= HAVE_METADATA)
> >          return m_player->duration();
> 
> While this is a sensible change, it shouldn't be necessary to prevent this
> crash because m_readyState should be HAVE_NOTHING if m_player is NULL. I
> haven't looked, but I suspect there are more problems waiting to be discovered
> when the the player has been destroyed and the state hasn't been reset.Can you
> please reset m_readyState and m_networkState to their default values when the
> player is destroyed?

The spec actually doesn't say that we should reset m_readyState to default
value, like HAVE_NOTHING. It specifies how m_networkState should be assigned
though. Though I think setting m_readyState to HAVE_NOTHING is sensible, but i
wasn't sure. Any thoughts?

> 
> 
> > Index: LayoutTests/http/tests/media/video-cancel-load.html
> 
> I would *really* like to see this test reformatted to have some structure (text
> in the <body>, <script> in <head> etc). I know that many layout tests do not do
> this, but I find that it is often much easier to understand what a test is
> supposed to do if has some structure.

Will do.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.