The JSC patch that changed text offsets to start from the open paren of the parameter list instead of the open brace of the function body broke the type profiler's protocol.
Created attachment 257626 [details] patch
Comment on attachment 257626 [details] patch View in context: https://bugs.webkit.org/attachment.cgi?id=257626&action=review > Source/WebInspectorUI/UserInterface/Models/ScriptSyntaxTree.js:179 > - divot: node.body.range[0] > + divot: node.isGetterOrSetter ? node.getterOrSetterRange[0] : node.range[0] // "f" in function, "s" in set, "g" in get, first letter in any method name for classes. Is this compatible with iOS 9 (a soon to be considered Legacy backend that wouldn't have had your recent backend changes). If not, maybe we should make this a function so that it would be easier to read. Something like: function functionReturnDivot(node) { // COMPATIBILITY (iOS 9): Legacy Backends something something something. // FIXME: Need a better way to determine backend versions. Using DOM.pseudoElement because that was added after iOS 9. if (!DOMAgent.hasEvent("pseudoElementAdded")) return node.body.range[0]; // "f" in function, "s" in set, "g" in get, first letter in any method name for classes. return node.isGetterOrSetter ? node.getterOrSetterRange[0] : node.range[0]; } Would be good to have a frontend test for this so it won't regress again in the future. But I won't make you write one now.
Created attachment 259178 [details] patch
Comment on attachment 259178 [details] patch View in context: https://bugs.webkit.org/attachment.cgi?id=259178&action=review > Source/WebInspectorUI/ChangeLog:8 > + Bug #145995 changed the starting offset of a function to Nit: We typically use the form <https://webkit.org/b/xxxxxx> when referencing bugs in the WebInspector change log.
Comment on attachment 259178 [details] patch r=me
Comment on attachment 259178 [details] patch Clearing flags on attachment: 259178 Committed r188549: <http://trac.webkit.org/changeset/188549>
All reviewed patches have been landed. Closing bug.