rdar://problem/35891328
Created attachment 328632 [details] Patch
Comment on attachment 328632 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=328632&action=review > Source/WebCore/ChangeLog:25 > + lists are tend to stay small. those lists tend (strikeout are)
Created attachment 328634 [details] Patch
Comment on attachment 328634 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=328634&action=review > Source/WebCore/accessibility/AXObjectCache.cpp:2729 > +static void filterForRemoval(const ListHashSet<T>& list, const Document& document, HashSet<Node*>& nodesToRemove) Why do we have a typename here for *list* but *nodesToRemove* is limited to <Node*>? This won't work if T is not a derived class of Node, right?
(In reply to Nan Wang from comment #4) > Comment on attachment 328634 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=328634&action=review > > > Source/WebCore/accessibility/AXObjectCache.cpp:2729 > > +static void filterForRemoval(const ListHashSet<T>& list, const Document& document, HashSet<Node*>& nodesToRemove) > > Why do we have a typename here for *list* but *nodesToRemove* is limited to > <Node*>? > This won't work if T is not a derived class of Node, right? ::remove() takes a Node& so nodesToRemove can't have non-Node based objects.
Comment on attachment 328634 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=328634&action=review >>> Source/WebCore/accessibility/AXObjectCache.cpp:2729 >>> +static void filterForRemoval(const ListHashSet<T>& list, const Document& document, HashSet<Node*>& nodesToRemove) >> >> Why do we have a typename here for *list* but *nodesToRemove* is limited to <Node*>? >> This won't work if T is not a derived class of Node, right? > > ::remove() takes a Node& so nodesToRemove can't have non-Node based objects. I see. But list has to contain only node based objects right? Why is the template necessary?
(In reply to Nan Wang from comment #6) > Comment on attachment 328634 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=328634&action=review > > >>> Source/WebCore/accessibility/AXObjectCache.cpp:2729 > >>> +static void filterForRemoval(const ListHashSet<T>& list, const Document& document, HashSet<Node*>& nodesToRemove) > >> > >> Why do we have a typename here for *list* but *nodesToRemove* is limited to <Node*>? > >> This won't work if T is not a derived class of Node, right? > > > > ::remove() takes a Node& so nodesToRemove can't have non-Node based objects. > > I see. But list has to contain only node based objects right? Why is the > template necessary? Correct. All incoming lists have to contain Node based object. However, while you can pass in an Element* when the function expects Node*, you can't pass in ListHashSet<Element*> for const ListHashSet<Node*>.
(In reply to zalan from comment #7) > (In reply to Nan Wang from comment #6) > > Comment on attachment 328634 [details] > > Patch > > > > View in context: > > https://bugs.webkit.org/attachment.cgi?id=328634&action=review > > > > >>> Source/WebCore/accessibility/AXObjectCache.cpp:2729 > > >>> +static void filterForRemoval(const ListHashSet<T>& list, const Document& document, HashSet<Node*>& nodesToRemove) > > >> > > >> Why do we have a typename here for *list* but *nodesToRemove* is limited to <Node*>? > > >> This won't work if T is not a derived class of Node, right? > > > > > > ::remove() takes a Node& so nodesToRemove can't have non-Node based objects. > > > > I see. But list has to contain only node based objects right? Why is the > > template necessary? > Correct. All incoming lists have to contain Node based object. However, > while you can pass in an Element* when the function expects Node*, you can't > pass in ListHashSet<Element*> for const ListHashSet<Node*>. I thought it's possible to pass in ListHashSet that holds the derived class objects. But ok then.
Comment on attachment 328634 [details] Patch Clearing flags on attachment: 328634 Committed r225613: <https://trac.webkit.org/changeset/225613>
All reviewed patches have been landed. Closing bug.