Load http://code.tutsplus.com/tutorials/say-hello-to-webkit-filters--net-23318 with the Inspector open. Didn't do anything special. TypeError: null is not an object (evaluating 'a.match(chunk)') comparePropertyDescriptors@file:///Users/Timothy/Work/Safari-TOT.git/OpenSource/WebKitBuild/Release/WebInspectorUI.framework/Resources/Views/ObjectTreeView.js:134:29 sort@[native code] _updateProperties@file:///Users/Timothy/Work/Safari-TOT.git/OpenSource/WebKitBuild/Release/WebInspectorUI.framework/Resources/Views/ObjectTreeView.js:287:24 _updateChildren@file:///Users/Timothy/Work/Safari-TOT.git/OpenSource/WebKitBuild/Release/WebInspectorUI.framework/Resources/Views/ObjectTreeView.js:257:21 _updateChildren@[native code] _getPropertyDescriptorsResolver@file:///Users/Timothy/Work/Safari-TOT.git/OpenSource/WebKitBuild/Release/WebInspectorUI.framework/Resources/Protocol/RemoteObject.js:510:17 _getPropertyDescriptorsResolver@[native code] _dispatchCallback@file:///Users/Timothy/Work/Safari-TOT.git/OpenSource/WebKitBuild/Release/WebInspectorUI.framework/Resources/Protocol/InspectorBackend.js:181:31 dispatch@file:///Users/Timothy/Work/Safari-TOT.git/OpenSource/WebKitBuild/Release/WebInspectorUI.framework/Resources/Protocol/InspectorBackend.js:86:35 dispatchNextQueuedMessageFromBackend@file:///Users/Timothy/Work/Safari-TOT.git/OpenSource/WebKitBuild/Release/WebInspectorUI.framework/Resources/Protocol/MessageDispatcher.js:42:34 file:///Users/Timothy/Work/Safari-TOT.git/OpenSource/WebKitBuild/Release/WebInspectorUI.framework/Resources/Controllers/FormatterContentBuilder.js:156:23: CONSOLE ERROR
<rdar://problem/20721693>
I don't understand how this is possible. Here is the code, annotated with some comments. ... // By this point, a and b should be strings. If they aren't something else went wrong in surrounding code. // 1. Rule out equivalent names. if (a === b) return 0; while (diff === 0) { // 2. Rule out one being shorter then the other. if (!a && b) return -1; if (!b && a) return 1; chunka = a.match(chunk)[0]; chunkb = b.match(chunk)[0]; anum = !isNaN(chunka); bnum = !isNaN(chunkb); // 3. Number versus non number cases. if (anum && !bnum) return -1; if (bnum && !anum) return 1; // 4. Number differences, if the same we continue. if (anum && bnum) { diff = chunka - chunkb; if (diff === 0 && chunka.length !== chunkb.length) { if (!+chunka && !+chunkb) // chunks are strings of all 0s (special case) return chunka.length - chunkb.length; else return chunkb.length - chunka.length; } } // 5. Different strings else if (chunka !== chunkb) return (chunka < chunkb) ? -1 : 1; // 6. Proceed to the next part of the string. a = a.substring(chunka.length); b = b.substring(chunkb.length); } ... For us to have encountered a.match(chunk) and have "a" be null then, I think something must have gone wrong. I believe Geoff just re-wrote Array.prototype.sort, so maybe that is related?
Tim, it looks like Geoff's sort changes (r183288) got rolled out a bit later (r183308). My ToT (r183442) was not encountering the issue. Do you know what r### your JavaScriptCore was?
I had Geoff's change at the time. Rebuilding to see if it is better.
Looks fine now.
I see this in TOT, and in 183568, just before the Array.prototype.sort patch re-landed: Uncaught exception in inspector page while handling event Network.responseReceived: (2) TypeError: undefined is not an object (evaluating 'b.resource.type') "compareResourceTreeElements@Main.js:6957:63 compareFolderAndResourceTreeElements@Main.js:6968:114 value@Main.js:516:107 insertionIndexForObjectInListSortedByFunction@Main.js:535:96 descendantResourceTreeElementTypeDidChange@Main.js:6825:240 callFirstAncestorFunction@Main.js:6238:130 _typeDidChange@Main.js:6992:112 dispatch@Main.js:100:158 dispatchEventToListeners@Main.js:102:9 updateForResponse@Main.js:3576:30 resourceRequestDidReceiveResponse@Main.js:14444:51 responseReceived@Main.js:797:69 dispatchEvent@Main.js:723:34 _dispatchEvent@Main.js:687:55 dispatch@Main.js:651:20 dispatchNextQueuedMessageFromBackend@Main.js:786:32" _dispatchEventMain.js:687:105 dispatchMain.js:651 dispatchNextQueuedMessageFromBackend
Geoff, was that loading the same page linked above?
(In reply to comment #7) > Geoff, was that loading the same page linked above? Yeah, same page.
Seems to be fine.(In reply to comment #8) > (In reply to comment #7) > > Geoff, was that loading the same page linked above? > > Yeah, same page. Filed new bug about that. Bug 147451.