[Webkit-unassigned] [Bug 112521] ScriptProcessorNode is garbage collected while still active if unreachable

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=112521


Russell McClellan <russell.mcclellan at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
         Resolution|FIXED                       |




--- Comment #13 from Russell McClellan <russell.mcclellan at gmail.com>  2013-03-17 23:54:40 PST ---
(In reply to comment #11)
> Has there been any verification that garbage collections is still happening correctly for dynamic nodes such as AudioBufferSourceNode?  For example in the granular demo, is the gc still happening correctly:
> http://chromium.googlecode.com/svn/trunk/samples/audio/granular.html
> 
> I'd be more comfortable if we could actually manually test that the gc is indeed happening in these cases.

Yes, I added some printfs to audiobuffersourcenode and tested that the C++ side objects were getting freed before I submitted it.  I just double-checked with granular.html and it seemed like garbage collection was happening correctly (that is, things were getting deleted at the same rate they were created)

> I'm not saying the patch is incorrect, but we should be extremely careful to test these types of things.

Yeah, I couldn't think of a great automated test for detecting that something was deleted correctly - I did try my best to test everything manually.

My original patch only affected ScriptProcessorNodes, but I changed it on Kentaro's recommendation (I agree with him that "pending activity" is better because the logic isn't duplicated between chrome and safari).  One point in favor of the current patch is that if audionodes ever become eventlisteners they will need a scheme like this or else they will have the same problem as ScriptProcessorNodes. (i.e., their event listeners will get deleted even though they are supposed to fire in the future).

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