[v8] Stop using an IDL to generate V8MutationCallback
Created attachment 181352 [details] Patch
This patch seems wrong, now we have manual code to put the callback into a hidden property that's exact duplication in both MutationObserver and WebAudio. Instead I want to just use that IDL attribute for web audio and other things with the same leak.
(In reply to comment #2) > This patch seems wrong, now we have manual code to put the callback into a hidden property that's exact duplication in both MutationObserver and WebAudio. Instead I want to just use that IDL attribute for web audio and other things with the same leak. Actually I'm wrong, this patch is fine!
Comment on attachment 181352 [details] Patch Ok. This patch makes me a bit sad because I'd like to have less custom code. I guess we can try to generate this code again if/when we have another callback like the mutation callback.
(In reply to comment #4) > (From update of attachment 181352 [details]) > Ok. This patch makes me a bit sad because I'd like to have less custom code. I guess we can try to generate this code again if/when we have another callback like the mutation callback. I'm only the tiniest bit sad. With only a single implementation, I'd rather have C++ code in source files than in perl. It'll be easy enough to move this back to the code generator if/when it makes sense.
Comment on attachment 181352 [details] Patch Clearing flags on attachment: 181352 Committed r138841: <http://trac.webkit.org/changeset/138841>
All reviewed patches have been landed. Closing bug.