On Sep 13, 2013, at 1:32 PM, Robert O'Callahan <robert@ocallahan.org> wrote: > On Fri, Sep 13, 2013 at 1:31 PM, Robert O'Callahan <robert@ocallahan.org> wrote: > On Fri, Sep 13, 2013 at 1:26 PM, Jussi Kalliokoski <jussi.kalliokoski@gmail.com> wrote: > I don't really care about using disconnect as pause, it's not meant for that and it's not a good solution for that. What I care about is what the use case / justification for having this weirdness is? A nice-to-have (not sure this is even a nice-to-have) doesn't really warrant introducing hard to debug memory leaks. I believe a lot of developers will expect that if they disconnect a ScriptProcessorNode and lose references to it, it gets GCed and doesn't burn cycles and / or eat away the performance of the graph anymore. This is similar to disconnecting a DOM node from the graph. If it has no references, it gets garbage collected and no more events are fired. > > That's not true for DOM nodes in general. For example, a completely disconnected HTML media element that's playing some media resource continues to fire events. > > I also think it would be bizarre for a developer to connect some audio node subgraph to the input of a ScriptProcessorNode for analysis, but not get events unless the output of that ScriptProcessorNode is connected to the DestinationNode. > That seems like a valid use case for an OfflineAudioContext. Why should that developer be rate-limited to 1 second-per-second if they wanted to analyze an entirely disconnected subgraph? -Jer
Received on Friday, 13 September 2013 20:34:42 UTC