This bug is also in Radar as <rdar://4258008> When we load content that ends up using the QTPlugin (or any plugin?), we cancel the load since WebKit plugins handle their own loading. This cancel is treated as an error, using WebKitErrorPlugInWillHandleLoad. We should figure out a way to provide status information for the load, or at the very least, not treat this cancel as an error.
Created attachment 6389 [details] example case To reproduce, download the archive and extract the two contained files to the same directory: qt-frame-test.html and qt.mov. Open qt-frame-test.html in Safari. Wait for the movie to start playing. Note that Safari's status bar and Activity window both list one error, though no details are shown anywhere.
Fixing this completely might require changes in both WebKit and Safari. Any change in WebKit that depends on a client change would have to be staged so that it continues to work no worse than it does now with old clients, while working better for new clients.
When I open qt-frame-test.html from the "example case" in WebKit r12753 (built locally) or in nightly WebKit-SVN-r12751.dmg, no movie appears in Safari and these messages are printed to the console (or written to console.log): Safari(3812,0xa000ed68) malloc: *** error for object 0x13839320: double free Safari(3812,0xa000ed68) malloc: *** set a breakpoint in szone_error to debug Is this another bug? Should it be filed in Radar or Bugzilla? I'm on Mac OS X 10.4.4 (Build 8G32) with Safari 2.0.3 (417.8).
(In reply to comment #3) > Safari(3812,0xa000ed68) malloc: *** error for object 0x13839320: double free > Safari(3812,0xa000ed68) malloc: *** set a breakpoint in szone_error to debug > > Is this another bug? Should it be filed in Radar or Bugzilla? > > I'm on Mac OS X 10.4.4 (Build 8G32) with Safari 2.0.3 (417.8). This happens on Safari 2.0.3 (417.8) WITHOUT an updated WebKit. Is this a regression? Safari(3832,0xa000ed68) malloc: *** error for object 0x577e2c0: double free Safari(3832,0xa000ed68) malloc: *** set a breakpoint in szone_error to debug
Hmm, I'm not seeing that, but it does sound like a problem.
(In reply to comment #4) > This happens on Safari 2.0.3 (417.8) WITHOUT an updated WebKit. Is this a > regression? > > Safari(3832,0xa000ed68) malloc: *** error for object 0x577e2c0: double free > Safari(3832,0xa000ed68) malloc: *** set a breakpoint in szone_error to debug Filed <rdar://problem/4442081> for this issue.
Ignore Comment #3 through Comment #7. These are a problem with my own PB G4, and are not related to this bug.