Summary: | [V8] NodeList wrappers are not kept alive as needed | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Erik Arvidsson <arv> | ||||||||
Component: | New Bugs | Assignee: | Erik Arvidsson <arv> | ||||||||
Status: | RESOLVED FIXED | ||||||||||
Severity: | Normal | CC: | abarth, haraken, japhet, jochen, ojan, rniwa, webkit.review.bot | ||||||||
Priority: | P2 | ||||||||||
Version: | 528+ (Nightly build) | ||||||||||
Hardware: | Unspecified | ||||||||||
OS: | Unspecified | ||||||||||
Attachments: |
|
Description
Erik Arvidsson
2012-06-28 11:55:16 PDT
Created attachment 149982 [details]
Patch
Comment on attachment 149982 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=149982&action=review > Source/WebCore/bindings/v8/custom/V8NodeListCustom.cpp:70 > + v8::Persistent<v8::Object> ownerWrapper = store->domNodeMap().get(owner); > + if (!ownerWrapper.IsEmpty()) { Does this over the case of getElementsByTagName? For that case, we probably don't want to add an implicit reference, right? Or does that actually always return the same object? Created attachment 150015 [details]
Patch
This now tests all DynamicNodeLists (except MicroDataItemList because it is not built by default). The V8 FAIL is due to 90208 Created attachment 150025 [details]
Patch for landing
Comment on attachment 150025 [details] Patch for landing Clearing flags on attachment: 150025 Committed r121496: <http://trac.webkit.org/changeset/121496> All reviewed patches have been landed. Closing bug. |