- From: Chris Rogers <crogers@google.com>
- Date: Tue, 19 Mar 2013 12:37:57 -0700
- To: Jussi Kalliokoski <jussi.kalliokoski@gmail.com>
- Cc: Srikumar Subramanian <srikumarks@gmail.com>, Peter van der Noord <peterdunord@gmail.com>, "public-audio@w3.org" <public-audio@w3.org>
- Message-ID: <CA+EzO0m3YsM=oXnoTX97iC39rLo+WcdY7YEWbxVXdDWo37YoSg@mail.gmail.com>
On Tue, Mar 19, 2013 at 12:33 PM, Jussi Kalliokoski < jussi.kalliokoski@gmail.com> wrote: > The bug against Chromium can be found here: > https://code.google.com/p/chromium/issues/detail?id=82795 > > Actually this reminds me... Is this actually a bug by the spec? I don't > think the current spec has any special rules for the ScriptProcessorNode > GC, so following the AudioNode garbage collection information in the spec, > it looks like ScriptProcessorNodes shouldn't be kept around if there aren't > any references to them. This has been a major source of confusion and I > think we should change this, and in general we should treat > ScriptProcessorNodes that have zero inputs as the sources that they are, > and a zero-input ScriptProcessorNode should be kept from garbage collection > as long as it's connected directly or indirectly to an active sink. > Jussi, it's not the node itself which is getting garbage collected, but the EventListener... > > Cheers, > Jussi > > On Tue, Mar 19, 2013 at 3:51 PM, Srikumar Subramanian < > srikumarks@gmail.com> wrote: > >> This is a long standing bug in the webkit implementation with a known >> workaround. If you keep around a *global* reference (in the window object >> itself) to either the onaudioprocess function or the script processor node >> itself, the arbitrary stopping won't happen. >> >> Best, >> -Kumar >> >> On 19-Mar-2013, at 4:40 PM, Peter van der Noord <peterdunord@gmail.com> >> wrote: >> >> I may be doing something wrong here, it's been a while since i created a >> script-processor node, but my event-handler for a scriptprocessor seems to >> stop firing pretty quickly. >> >> Fiddle here: http://jsfiddle.net/7LZAJ/ >> >> Strangely, the fiddle sometimes runs fine the first time, but when i >> refresh after that it does what happens locally here as well: the >> eventhandler does not get called anymore (and the remaining noisebuffer is >> repeated). The writing of the noise into the buffer has nothing to do with >> it, if my eventhandler only logs something to the console it will just >> stop shortly after starting (always after the same number of repetitions). >> >> Am i overlooking something? >> >> >> Peter >> >> >
Received on Tuesday, 19 March 2013 19:38:25 UTC